Documente Academic
Documente Profesional
Documente Cultură
HYDRODYNAMICS
AND WATER WAVES
HYDR ODYN AMIC A
SIVE
DE VIRIBUS ET MOTIBUS FLUIDORUM
COMMENTA RII
HYDRODYNAMICS
AND WATER WAVES
Bernard Le Mehaute
PART TWO
PART THREE
---
space, is also a vector quantity. The components of grad p / /
along the three coordinate axes OX, OY, OZ, are given by
/ / .& STREAMLINE
the derivative of p with respect to x, y, z, respectively; i.e.,
/ ~
8pj8x, 8pj8y, 8pj8z. / /
ofV along the three Cartesian coordinate axes OX, OY, OZ, I ~ / /
I
are u, v, and w, respectively. If i, j, k are unit vectors along I I /
the axes OX, OY, OZ respectively, then: V = iu + jv + kw.
Since the system of reference is rectangular, the magnitude Figure 1-2 Streamlines observed by short-exposure
of the velocity is given by V = [u 2 + v2 + w2 2 V is a r1 photography of various particles.
scalar quantity and therefore completely defined by its
magnitude, like the pressure p. V is a vector quantity and is
specified by its direction and magnitude. Since V and its This is the mathematical definition of a streamline. These
components u, v, and w are functions of the space co- equalities express the fact that the velocity is tangential to
ordinates of A and the timet, they can be written in the form the displacement of the particle at time t 0 . Figure 1-3
V(x,y,z,t). illustrates this fact in the case of a two-dimensional motion.
In this case dxju = dyjv, which implies v dx - u dy = 0.
1-2.2 Definitions Streamlines do not cross, except at point of theoretically
1-2.2.1 The displacement dS of a fluid particle is defined infinite velocity (see Figs. 11-6 and 11-7) and at stagnation
by the vector equation, dS = V dt, which is valid for both and separation points of a body where the velocity is zero.
magnitude and direction. This equation may be written Fixed solid boundaries and steady free surfaces are stream-
more specifically in terms of the displacements in each of lines. Moving boundaries, such as propeller blades, and
the three Cartesian coordinate directions as follows: unsteady free surfaces are not streamlines.
dx = u dt
1-2.2.3 The path of a specific particle of the fluid is
dy = v dt defined by its position as a function of time. It may be
dz = w dt
1-2.2.2 A streamline is defined as a line which is tangential Figure 1-3 Definition of a streamline in a
at every point to the velocity vector at a given time t 0 . two-dimensional motion.
A device for visualizing streamlines is to imagine a number y
of small bright particles distributed at random in the fluid,
and then to photograph them with a short exposure v
(Fig. 1-2). Every particle photographs as a small line
segment. Each line which is drawn tangentially to these
small segments is a streamline.
v 1
dy
At time t 0 , the equations dx = u dt, dy = v dt, and
dz = w dt become:
dx dy dz
0~--~=d=x::~:---~x
u(x,y,z,t 0 ) v(x,y,z,t 0 ) w(x,y,z,t 0 ) u
Part 1 : Establishing the Basic Equations that Govern Flow Motion
1-2.2.5 An elementary flow channel bounded by an streakline; if the successive location of a neutrally buoyant
infinite number of streamlines crossing a closed curve small ball are determined, a particle path can be traced;
is known as a stream tube (Fig. 1-5). finally, if a large number of short threads are attached to a
1-2.3 Steady and Unsteady Flow body, the instantaneous direction of these threads will
yield a streamline pattern. All these methods are commonly
1-2.3.1 For steady flows defined by time-independent used in fluid flow experimental studies.
quantities, streamlines, streaklines, and particle paths Streamlines, paths, streaklines, and stream tubes are
are identical. However, for unsteady or time-dependent different in unsteady flow, that is, flow changing with
flows, these lines are different and a clear understanding respect to time. Turbulent flow is always an unsteady flow;
of their generation is necessary to properly interpret the however, it will be seen in that case that the mean motion
results of a given experiment. For example, if dye is injected with respect to time of a turbulent flow may be considered
at a given point of a fluid flow, the dye pattern will be a as steady. Then streamlines, paths, and streaklines of the
mean motion are the same (see Chapter 7). Figures 1-6 and
1-7 illustrate these definitions in some cases of unsteady
Figure 1-4 Streakline obtained by instantaneous
photography of various particles
motion.
coming from the same point.
1-2.3.2 In some cases of unsteady flow (a body moving
at constant velocity in a still fluid, a steady wave profile
such as those due to a periodic wave or a solitary wave) it is
possible to transform an unsteady motion into a steady
motion relative to a coordinate system which moves with
the body or the wave velocity. The construction of a steady
pattern is then obtained by subtracting the velocity of the
body from the velocity of the fluid. This is the Galilean
transformation. Steady streamlines can then be defined
relative to a moving observer who travels with the body or
with the wave (see Fig. 1-8).
Chapter 1 : Basic Concepts and Principles
WAVE TRAVEL 7
Figure 1-6
Periodic gravity wave in deep water.
1-3 Methods of Study compressible fluid, densities and temperatures are also
given in terms of the original position and the elapsed time.
The motion of a fluid can be studied either by the method
If the initial position of a given particle at time t 0 is
of Lagrange or the method of Euler.
x 0 , y 0 , z0 , a Lagrangian system of equations gives the
1-3.1 Lagrangian Method position x, y, z, at the instant t as:
The Lagrangian method may be used to answer the X= F 1(x 0 ,y0 ,z 0 ,t- t 0 )
question: What occurs to a given particle of fluid as it y = F 2 (x 0 ,y0 ,z0 ,t- t 0 )
moves along its own path? This method consists offollowing z = F 3 (x 0 ,y0 ,z 0 ,t - t 0 )
the fluid particles during the course of time and giving the
paths, velocities, and pressures in terms of the original In practice this method is seldom used in hydrodynamics.
position of the particles and the time elapsed since the Lagrangian coordinates are, however, often used in theories
particles occupied their original position. In the case of a relative to periodical gravity waves. The velocity and
WIND
... SMOKE
Figure 1-7
Smoke in the wind.
Part I : Establishing the Basic Equations that Govern Flow Motion
8 Figure 1-8 (Top) Streamlines, paths, streaklines for a acceleration components at point (x 0 ,y0 ,z 0 ) are then
steady flow around a .fixed body. obtained by a simple partial differentiation with respect
(Middle) Streamlines, paths for an unsteady to time, such that
flow around a body moving at constant
velocity in a stillfluid.
(Bottom) Vectorial relationship between the
U =
OX
7ft
I
xo,yo,zo
two kinds of motion: Galilean
transformation.
oz I
W = Ot Xo,yo,zo
V = F(x,y,z,t)
then
u f 1(x,y,z,t)
=
v = f 2 (x,y,z,t)
w = f 3 (x,y,z,t)
and
p = F 1(x,y,z,t)
Thus,
dx dz
k(Hj2)emz cos (kt 0 - mx) -k(H/2)emz sin (kt 0 - mx)
This is the equation of a circle of radius (Hj2)emzo. It is seen
or that the paths are circular and the radius tends to zero as
z 0 --+- oo. It will be seen in the linear wave theory that, at a
dz =-tan (kt 0 - mx) dx first approximation, one has X; ~ x 0 , z; ~ z 0 (Section 16-1),
and x 0 , z 0 can be considered the location of the fluid particle
If t 0 is taken as 0, this equation becomes: at rest.
ro being vectorial. If V is expressed by its components u, v, fluid), it gives a relationship between the components of V
and w, four scalar or ordinary equations are necessary. and the space coordinates, which are x, y, z. The equation
In free surface flow problems, the free surface elevation of continuity then becomes
11(x,y,z,t) around the still water level, or the water depth
h(x,y,z,t), is unknown and a kinematic condition is also au+ av + aw = 0
required. However, in that case the pressure p is known and ax ay az
in general is equal to the atmospheric pressure. as it is demonstrated in Section 3-2.
For gases, two more unknowns need to be considered, It will be seen that V may be found in some cases of flow
namely, the density p and the absolute temperature T. under pressure, independent of the absolute value for p, from
Hence, to solve problems in the most general cases of fluid the principle of continuity alone, but p will always be a
mechanics, four equations are necessary. If V is expressed function of V except at the free surface.
by u, v, and w, then six ordinary equations are needed.
In hydrodynamics, basic equations are given by the 1-4.3 The Momentum Principle
physical principles of continuity and conservation of
momentum. The equation of state and the principle of the The momentum principle expresses the relationship
conservation of energy must be added in the case of com- between the applied forces F on a unit volume of matter of
pressible fluid. density p and the inertia forces d(pV)/dt of this unit volume
The reduction of a problem to such a small number of of matter in motion. The inertia forces are due to the natural
variables (2 in hydrodynamics and 4 in gas dynamics), does tendency of bodies to resist any change in their motion.
not occur for trivial reasons, but as a result of several It is Newton's first law that "every body continues in its
important arguments and assumptions. A number of state of rest or uniform motion via a straight line unless
phenomenological functions are assumed to be known. For it is compelled by an external force to change that state."
example, it is assumed that the fluid is Newtonian and The well-known Newtonian relationship is derived from
either perfect or viscous, which defines the stress tensor. his second law: "The rate of change of momentum is
The fluid obeys Fourier's law of conduction. Also, a proportional to the applied force and takes place in the
number of coefficients, such as heat conductivity, specific direction in which the force acts." F = d(mV)/dt.
heat, and viscosity, are supposed to be known functions In fluid mechanics this equation takes particular forms
of the other unknown variables, such as density and/or which take into account the fact that the fluid particle
temperature. may be deformed. These equations will be studied in detail.
For an incompressible fluid, the integration of the momen-
tum equation with respect to distance gives an equality
1-4.2 Principle of Continuity of work and energy, expressing a form of the conservation
The continuity principle expresses the conservation of of energy principle.
matter, i.e., fluid matter in a given space cannot be created If Vis expressed by u, v, w, then Newton's second law
or destroyed. In the case of an incompressible homogeneous has to be expressed along the three coordinate axes. This
fluid, the principle of continuity is expressed by the conserva- gives the three equations
tion of volume, except in the special case of cavitation where
partial voids appear.
The continuity principle gives a relationship between
the velocity V, the density p, and the space coordinates where pis assumed constant and Fx, FY, Fz are the com-
and time. If p is constant (in the case of an incompressible ponents of F along the three coordinate axes, respectively.
Chapter 1 : Basic Concepts and Principles
op op op op
dp = - dx +- dy +- dz +- dt = 0
ox dy az at
dp = ap + op dx + ap dy + op dz = 0
dt ot ax dt oy dt oz dt
op op ap op
-+u-+v-+w-=0
at ox oy az Figure 1-9 Uniform flow in a rectangular channel.
_1 2A
,_ z 13
I I I
I I
Figure 1-10 A piston wave paddle gives a movable boundary Figure 1-12 Periodic gravity wave in infinite
condition.
depth: V--> 0 when z--> - oo.
of the boundary itself. The other component follows the surface. Hence, the periodic gravity wave theory in infinite
corresponding boundary motion component for a real depth is based on the boundary condition V -+ 0 when the
fluid only. distance from the free surface tends to infinity: z -+ - oo
If F(x,y,z,t) = constant is the equation of the movable (Fig. 1-12).
boundary, the following boundary condition expresses
the fact that the fluid remains at the boundary:
PROBLEMS
aF aF aF aF
-+u-+v-+w-=0
at ax ay az 1.1 Consider a two-dimensional flow motion defined by
the velocity components:
1-4.6.5 An infinite distance can give a boundary condition u =A+ Bt v= C
if the motion tends to a well known value far from the where A, B, and C are constant parameters. Demonstrate
studied space. For example, consider the diagram shown that the streamlines are straight lines and that the particle
in Fig. 1-11. The motion is known at infinity and can be paths are parabolas.
written (as far as friction effect is negligible) V = constant
for x-+ oo. 1.2 A disk of radius R rolls without slipping on a horizontal
As another example, it is well known that the motion plane at a constant angular velocity k. Demonstrate that
of swell in deep water is limited to a zone near the free the "streamlines" are circular and that the paths are
trochoidal.
Figure 1-11 Flow in a pipe past a diaphragm: 1.3 Consider a fixed cylinder in a uniform current of
V = V0 when x --> oo. constant velocity. It will be assumed that there is no separa-
tion. Sketch the streamlines, the paths, and the streaklines
intuitively. Consider now a cylinder moving at constant
velocity in still water, and sketch the streamlines, the paths,
and the streaklines. Explain the differences between the
two cases, considered as a steady and an unsteady motion,
respectively.
Part 1 : Establishing the Basic Equations that Govern Flow Motion
I4 1.4 Devise a general graphical method of construction to period; and d is the water depth.) Find the equations for
determine a steady flow pattern around a body moving the streamlines, and sketch them. Assuming Ix - x 0 I
at constant speed in a still fluid from the pattern of stream- and lz- z0 1 are small, find the approximate equation
lines defined with respect to a fixed coordinates system for the particle paths. Sketch these paths.
and vice versa.
1.6 Express mathematically the boundary conditions for
1.5 A two-dimensional flow motion (linear periodic gravity
any kind of flow motion taking place between the bound-
waves in water depth, d) is defined in a Lagrangian system
aries defined by Fig. 1-13. A hinged paddle will be assumed
of coordinates by the equations:
to have a small sinusoidal motion of amplitude e at the
_ . H cosh m(d + z) . (k _ ) free surface.
x - x, + 2 . h d Sill t mx
sill m
H sinh m(d + z) 1.7 Consider a two-dimensional body moving at velocity
z = z; + -2 . h d cos (kt - mx) U in the negative X direction. The nose of this body can
Sill m
be defined by a curve such that y = x 1 i 3 , and u and v
where H is the wave height; m, k, and d are constants. are the components of velocity along the body. Establish
(m = 2njL; L is the wavelength; k = 2n/T, Tis the wave the relationship between u, v, U, and y.
1##-fr#$#/ll.f$/.f;!l.f$,
-- - 1/11/l/l//l/l/lll/(1111}
b c
.-
0
------
Figure 1-13
Wave basin, top and side view.
~ ~REE SURFA~
Chapter 1 : Basic Concepts and Principles
Then consider the case where the body is fixed and the 15
fluid is moving at a velocity U.
c c _ C1 Cz _ (oufoy) dy dt
2 3- J2 - J2
The rate of angular rotation is:
dr 1 au
dt 2 ay
Chapter 2: Motions of Fluid Elements: Rotational and Irrotational Flow
u + ~ dy \ B,
~y /-
r--
I
8,...-
________,c ____ -~{--\..:,;:8.:...'____,_;;,.-.,.;:-...,c,
I'
: ',
Figure 2-1
L
I
dy I
Elementary analysis of different kinds I
of motion of a fluid particle. I /
I,/
Ai, u dx _JD-----A, ~----~
Similarly, the rate of deformation would be found to be velocity components. The translation can be along a straight
equal to: line or a curved line.
~ (C3C1) _~au
at A1C3 2ay
If x, y, and z are the coordinates of A at time t, then
x + L\x, y + Lly, and z + Llz are the coordinates at time
t + M. The translatory motion is defined by the equations
In the general case, there are three major constituents of Llx = u Llt dx = u dt
particle motions and deformations. They are: Lly = v Llt or dy == v dt
1. The velocity components V(u,v,w): translation Llz = w Llt dz = w dt
2. The variation of velocity components in their own The flow of particles along parallel and straight stream-
direction: dilatation lines with a constant velocity (so-called uniform flow) is a
3. The variation of velocity components with respect to case of translatory motion only (Fig. 2-3).
a direction normal to their own direction: rotation The translatory motion may be defined more rigorously
and angular deformation as the motion of the center of the rectangular element
These three constituents are successively discussed in the
following sections. Figure 2-2 Translatory motion.
z
2-2 Translatory Motion
Consider the particle at the point A(x,y,z) at time t.
The point is a corner of a small rectangular element, the
edges of which are parallel to the three axes OX, OY, OZ,
respectively (Fig. 2-2). When the particle moves so that the
edges of the rectangular elements remain parallel to these
axes, and maintain a constant length, it is a translatory
motion only. This implies no space dependence of the y
Part I : Establishing the Basic Equations that Govern Flow Motion
L.
I
v bu I
u ut~ d~ I
I
I
'
L :~dxdl
~---'----~---------_J
A
_JD
Chapter 2: Motions of Fluid Elements: Rotational and Irrotational Flow
:l ; '" r
V x R = constant may be approximately applied where square element tend to remain parallel to their initial
bU
u+;-dy
0 y I I
- ' _ __,o_ - --7 o'
c r ---Tc_
Figure 2-7 I
I I
Angular or shear deformation. y ~dt I
by I I
I I
I
---
I
A L - - -- - -..J B
0~----~- x
Part 1 : Establishing the Basic Equations that Govern Flow Motion
20 positions during the angular deformation. When the bisectors change their direction, and there is either both
bisectors do not remain parallel to their initial positions, rotation and angular deformation, or rotation only (Fig.
the motion is said to be rotational. 2-8).
The difference (oujoy) - (ovjox) defines the rate of
rotation, and therefore, a two-dimensional irrotational
2-4 Rotation motion is defined mathematically by (oujoy) - (ovjox) = 0.
Although flow motions can be classified in various ways Angular deformation can be considered without rotation
according to some of their typical characteristics (such as when (oujoy) - (ovjox) = 0 and (oujoy) + (ovjox) # 0, and
laminar or turbulent, frictionless or viscous, with or without theoretically, rotation can exist without deformation when
friction, steady or unsteady), one of the most important (oujoy) - (ovjox) # 0 and (oujoy) + (ovjox) = 0. This case
divisions in hydrodynamics consists of considering whether is rare in practice, since rotation generally occurs with
a flow is rotational or irrotational. Hence, the abstract angular deformation in physical situations. A forced vortex,
concept of irrotationality is fully developed in the following such as that schematically shown on Fig. 2-9, is a rare case
sections. in which particles rotate without deformation. However,
this can be considered more as a special case of hydrostatics
2-4.1 Mathematical Definitions where the centrifugal force is added to gravity, rather than
For a two-dimensional motion, it has been shown that a real rotational flow.
the angular velocities of deformation are oujoy and ovjox.
The rotation of a particle is proportional to the difference 2-4.2 Velocity Potential Function:
between these components. Indeed, if oujoy = ovjox, Definition
there is angular deformation without rotation and the The concept of irrotational motion is very important in
bisectors do not rotate (Fig. 2-7). But if oujoy # ovjox, the hydrodynamics since many real flows are nearly irrotational.
I
I
'jl
Figure 2-8
Rotation and deformation.
SHEAR
DEFORMATION ROTATION ROTATION AND
WITHOUT ROTATION WITHOUT DEFORMATION DEFORMATION
6u 6v 6u
- ov ou - OV
ry - 6X = 0 dy 6X
;. 0 6y 6X
;. 0
6u 6v ;. 6u 6v 6u 6v ;.
6-y +6- 6-y + 6X YY + IX
0 0 0
X
Chapter 2: Motions of Fluid Elements: Rotational and Irrotational Flow
where i and j are the unit vectors along the X and Y axis, 21
respectively. The magnitude of the velocity becomes
Figure 2-12
Examples of rotational and irrotational motion.
Part I : Establishing the Basic Equations that Govern Flow Motion
24 IRROTATIONAL (Convergent)
VE NT
Figure 2-13
Examples of rotational and irrotational motion.
the Gerstner's theory on periodic gravity waves. In this rotational terms are taken into account exactly (see Section
theory the paths of fluid particles describe circles. The 17-1.4).
particles also rotate about themselves in the opposite
direction (Fig. 2-15). The results are expressed by an exact 2-5.3 Irrotational Solutions in Viscous
mathematical solution of the basic equations in which the Fluids
friction terms have been neglected, but in which the inertial One also finds dissipative motions, which are considered
as irrotational. For example, friction forces have a pre-
dominant effect on such phenomena as the damping of a
Figure 2-14 Effect of the boundary layer.
DIRECT ION OF
WAVE T RAVE L
- A I R ENT RAINMENT
Chapter 2: Motions of Fluid Elements: Rotational and Irrotational Flow
gravity wave though a filter and the flow through a porous fluid. The velocity gradient normal to streamlines can be 25
medium. However, in these cases, only the mean velocity large near the cylinder. The motion is still irrotational.
with respect to space is considered. The actual system of The motion of a free vortex is the same whether one
complicated rotational motions through the porous medium considers the fluid perfect or viscous. The solution to the
is studied as an average motion with respect to space which momentum equation for a perfect fluid (VR = constant)
is irrotational at low Reynolds number (see Chapter 9). makes the sum of all viscous terms of the momentum
Similarly, turbulent flow is strongly rotational but the equation equal to zero.
mean motion with respect to time may often be considered
as irrotational (see Chapter 8). 2-5.4 Energy Dissipation, Shear
It may also happen that the flow is irrotational when the Deformation, and Rotationality
sum of all the viscous terms which appear in the momentum The fact that the motion is rotational does not necessarily
equation equal zero, although each term individually is mean that it is dissipative. A motion is dissipative when
different from zero. Such kinds of motion are dissipative there are linear and/or angular deformations associated
and irrotational. A specific example of such a case is the with a non-negligible viscous coefficient. So an irrotational
motion generated by a circular cylinder rotating steadily free vortex can be dissipative, while a rotational forced
about its axis in an unbounded viscous incompressible vortex is not dissipative.
o'
y+vdt
Figure 2-16
Two-dimensional coordinate system.
y + dy
C:''
v
X
y ----- t
A:-u 9
ou
u -t-dx
I ox
I I
I I
~--------~-----------J----------------J-------~- X
0 )( x + dx x+udt
Part 1 : Establishing the Basic Equations that Govern Flow Motion
26
x + dx + u dt ou dx dt + -1 (ou
+ -ox - + -ov) dy dt - -1 (ov
- - -ou) dy dt
2 oy ox 2 ox oy
D' (2-2)
y + dy + v dt + ov dy dt + ~ (ou + ov) dx dt + ~ (ov - ou) dx dt
oy 2 oy ox 2 ox oy
'--,.--' "-.-' "-.-' '-.-' '-.-'
Initial Translation Dilatational Rate of angular Rate of
coordinates or linear deformation rotation
deformation
Angular or shear Rotation
deformation
A'{xy ++ vudtdt y
ov ov ov
+ dy + v dt + ( dx dx + oy dy + jjz dz dt
) (2-3)
(2-1)
D'{x + dx + (u + du) dt ow ow ow )
y + dy + (v + dv) dt z + dz + w dt + ( ox dx + oy dy + oz dz dt
Chapter 2: Motions of Fluid Elements: Rotational and Irrotational Flow
27
x + dx + u dt + -au dx dt + [1- (ov
ax
- + -au) dy + -1 (au
2 ax ay
- + -ow) dz + -1 (au
2 oz
- - -ow) dz
ax 2 oz ax
- -1 (av
- - -au) dy dt
2 ax oy
J
y + dy + v dt + -ov dy dt + [1- (ow
ay
- + -ov) dz + -1 (ov
2 oy az
- + -au) dx + -1 (av
2 ax
- - -au) dx
ay 2 ax ay
- -1 (ow
- - -ov) dz dt
2 ay az
J (2-4)
z + dz + w dt + c dz dt + (g dx +f dy) dt + (~ dy - 17 dx) dt
'--.-' ~ ~
Initial Translation Dilatational Angular Rotation
coordinates deformation deformation
Part 1 : Establishing the Basic Equations that Govern Flow Motion
28 which represents the vorticity of the fluid at a point. A The above equations would still hold, even if the velocity
three-dimensional irrotational motion is defined by ~ = 0, potential had been negative, so the velocity potential can
rJ = 0, and ( = 0; that is, also be defined by V = -grad .
Figure 2-17
In a thin .flow of water, rotation exists
only in vertical planes.
GLASS PLATE
Chapter 2: Motions of Fluid Elements: Rotational and Irrotational Flow
29
Figure 2-19
INTAKE
2.3 Determine the coefficients of dilatational and shear
deformation and rotation for a flow between two parallel
planes separated by a distance d = O.OlL. One of the
planes is assumed to be fixed, the other one moves at a
speed V = O.lLjt. The velocity distribution between the
INJECTION two planes is linear.
OF DYE
Figure 2-18 Examples of studies based on Stokes analogy. where J1 is the coefficient of viscosity, e is the distance
between the two planes, a: is a constant equal to the head
loss or decrease of pressure per unit length: a: = dpjdx.
PROBLEMS
2.2 Indicate the domains of Fig. 2-20, where the flow can be
considered as irrotational and the domains where the flow
is rotational. Give the reasons which prevail in your choice.
Part 1 : Establishing the Basic Equations that Govern Flow Motion
30 V is the velocity of one of the planes, the other one being turbulent flow through rocks are
assumed to be fixed.
Determine the coefficient of dilatational and shear op = Ku2
deformation and rotation as a function of y. Consider the ox
two cases where oc = 0 and V # 0 on one hand, and oc # 0
op = Kv2
and V = 0, on the other hand, as two particular cases,
and explain their significance.
ay
Demonstrate that such a flow is in general rotational.
2.5 Express the velocity components as a function of a 2.9 Demonstrate that at a given location in a two-dimen-
potential function in a cylindrical (r,lJ,z) and spherical sional flow, the value of angular rotation is independent of
(r,ci>,lJ) system of coordinates. the axis system of references, i.e.,
2.6 Derive the expression for irrotationality in polar system ! (av _ au) = ! (ov' _ ou')
of coordinates (r, lJ). The components of velocity are: 2 OX oy 2 ox' oy'
radial velocity v, tangential velocity v8 u' and v' being the velocity components along the x' axis
and the y' axis, respectively.
2.7 Consider two coaxial cylinders of radius R 1 and R 2
2.10 Demonstrate that
which are rotating at angular velocity w 1 and w 2 , respec-
tively. The velocity distribution of the fluid between these
two cylinders is given as a function of r by the expression
(R 1 < r < R 2 )
op = Ku
ox
op = Kv
dy
where p is the hydrostatic pressure. Demonstrate that such
a flow is irrotational. The equations for an average fully
Chapter 3 3-1 Elementary Relationships
31
dV =au= O
dx ax
The case of a two-dimensional motion of an incompres-
sible fluid is now given. The general case will be treated
later.
y
32 After a time dt, because of the change of density with
respect to time, the quantity of fluid mass becomes
p dx dy dz-
ap dt ) dx dy dz = -at
( p +at op dt dx dy dz
X (3-1)
0 ~--------------------~
Figure 3-1 Rectangular element of an On the other hand, if one takes into account the change in
incompressible fluid. velocity and in density with respect to space coordinates, the
quantity of fluid mass entering through the section ABCD
during a time dt, parallel to the OX axis, is the product pu
Since the total mass of fluid within the boundaries is times the area perpendicular to OX (ABCD) and the time dt.
constant, the total loss must be zero: a du + b dv = 0; Since ABCD = dy dz, the quantity of fluid mass entering is
that is, (du/b) + (dvja) = 0. pu dy dz dt. The derivative of u along AB and AD with
In the limit, when b and a approach zero, one obtains respect to dz and dy is of an infinitely small order and can be
(oujox) + (ovjoy) = 0. This differential form is permitted neglected. Now the quantity of fluid mass coming out
because of the assumption of a continuous fluid. It should
be noted that oujox and ovjoy are the rates oflinear deforma-
tion of a fluid particle; hence, in an incompressible fluid, Figure 3-2 Coordinate system for continuity
the total sum of linear deformation is nil, as has been pre- equation.
viously noted.
c G
Relationship ~-------- F
I
I
Consider a fixed volume of fluid of which the edges dx, I
I
dy, dz are parallel to the axes OX, OY, OZ, respectively A
(Fig. 3-2). The continuity relationship is obtained by
considering that the change of fluid mass inside the volume X
during the same interval of time through the section These continuity relationships can be written in a 33
EFGH is: shorter way as follows:
[ pu o(pu)
+ ---a;: dx
Jdy dz dt -op + d.IV pV
at
=
o
In the general case, both the density p and velocity u are or
assumed to be changed along dx. Hence the difference is
~ + p div V + V grad p = 0
[ pu
o(pu)
+ ---a;: dx
Jdy dz dt - pu dy dz dt =
o(pu)
---a;: dx dy dz dt
3-2.2 Physical Meaning and
Similarly, the difference due to the components of Approximations
motion parallel to the OY and OZ axes are, respectively, Consider, respectively, the three groups of terms:
o(pv) d d d d due to the difference of discharge
Tyxyzt op
across the sections BFGC and AEHD
at
(dx dz)
ou + ov +OW= 0 or div V = 0
ax oy az
Since [o(pu)]jox = p(oujox) + u(opjox), and similarly
for the terms o(pv)joy and o(pw)/oz, the continuity relation- When div V > 0, an expansion of the fluid is indicated,
ship becomes and conversely, div V < 0 signifies a compression.
op (au ov op op op
-+p -+-+- +u-+v-+w-=0
ow) 3-2.2.3 The third group of terms is proportional to the
at ax oy az ax oy az derivative of density with respect to the space coordinates
Part 1 : Establishing the Basic Equations that Govern Flow Motion
34 at a given time. This derivative is usually negligible in This phenomenon occurs when a large vanatwn of
comparison with the others. pressure results in the celerity of the wave exceeding the
For example, consider a unidimensional sinusoidal velocity of sound. A shock wave travels at a higher speed
acoustic wave. The density is given as a function of time than usual pressure waves such as acoustic sound or water
and distance by the relationship p = A sin (Ct + x). The hammer. Hence, when a supersonic flow or the effect of an
derivative of p with respect to time t is opjot = AC cos underwater explosion is studied, the derivative of the
(Ct + x), and with respect to the distance x along the OX density with respect to the distance has to be taken into
axis, opjox = A cos (Ct + x). Hence, account in the continuity relationship.
opfox
3-2.2.4 Table 3-1 summarizes these considerations.
op/ot c
Unless otherwise specified, incompressible fluids only are
where C is the wave velocity. considered in the rest of this text.
Since Cis usually large compared to the particle velocity
u, u(opjox) is usually negligible by comparison with opjot.
However, there is an exception to this rule: the case of 3-3 Some Particular Cases of the
shock waves where the variation of p with respect to space Continuity Relationship
is theoretically infinite at the front of the wave.
The continuity relationship is often used in other forms
Table 3-1 in hydraulics. These forms are not so general but more
adapted to integration for the phenomena to be studied.
Some examples of the different forms used are provided
Uniform (one-dimensional) au= 0
flow of an incompres- ax by the case of unsteady flow, mainly unidimensional either
sible fluid at a free surface (channel, river) or under pressure (pipe,
gallery).
Two-dimensional flow of an
incompressible fluid
An unsteady free surface flow resulting in a change oflevel
with respect to time and space and caused by gravity
Three-dimensional flow of an au + av + aw = O action is called a gravity wave. Some examples are flood
incompressible fluid ax ay az waves in a river, bore (translatory waves), tsunami waves
due to earthquake, tides, harbor oscillation, and seiche.
div V = 0
An unsteady flow under pressure resulting in a change of
pressure with respect to time and space, caused by the
Unsteady motion in a ap + p(au + av + aw) = 0
compressible fluid at at ax ay az pressure gradient is called a pressure wave. Two examples
usual speed (acoustic are water hammer and acoustic wave. Such gravity waves
wave, water hammer)
~~ + p div V = 0 and pressure waves are studied from special forms of the
continuity relationships. They are valid when the distri-
Unsteady motion in a ap + a(pu) + a(pv) + a(pw) = 0 bution of velocity in a cross section can be assumed to be a
compressible fluid at at ax ay az constant; then the gravity waves are called long waves.
high speed (shock wave) Here, the continuity relationship is obtained by stating
~ + div (pV) = 0 that the change of volume of water during the interval of
time dt between two finite cross sections separated by the
Chapter 3: The Continuity Principle
infinitely small distance dx is equal to the difference between Equating Equations 3-3 and 3-4 and dividing by dx dt, one 35
the influx and efflux from the considered volume during the obtains (ahjat) + (aqjax) = 0, (ahjat) + (ahujax) = 0, or
same interval of time
ah + u ah + h au = 0
at ax ax
3-3.1 Translatory Gravity Waves
Consider the volume defined by the two cross sections x 3-3.2 Irrotational Flow
and x + dx and the free surface at time t (Fig. 3-3). The If the density p is a constant, the continuity relationship
volume of influx, during a time dt, into the considered has been seen to be (aujax) + (avjay) + (awjaz) = 0.
volume at x is q dt or hu dt, where q is the discharge, h the In the case of irrotational motion, a velocity potential
depth, and u the horizontal velocity component. function has been defined by the relationships
The efflux out of the volume at x + dx is
a a a
[h J U=- V=- w=-
ax ay az
or u+~
a(hu) dx dt
Hence, introducing these expressions into the continuity
relationship yields
Hence the change of volume ~ V between these cross
sections x and x + dx is a difference az az az
ax 2 + ay 2 + az 2 = 0
aq ~V a(hu) dx dt (3-3)
~v = - -dxdt or = -
ax
which can be written '\1 2 = 0. This is the well-known
ax
Laplace equation which has been subjected to extensive
On the other hand, the volume at the time t is h dx, and research in mathematical physics.
at the time t + dt, since the free surface level changes the
3-3.3 Lagrangian System of Coordinates
volume is [h + (ahjat) dt] dx. Hence the change of volume
during time dt is: It has been explained in Chapter 1 that it is possible to
study problems in hydraulics either in Eulerian coordinates
ah or in Lagrangian coordinates (See Section 1-3.1). Since
~v = atdtdx (3-4)
this last system of coordinates is rarely used, the continuity
----t--
Figure 3-3 C:>h
1
h+bi dx I
Translatory wave.
I TRANSLATORY
I WAVE PROFILE
I
I
0 x tdx
Part I : Establishing the Basic Equations that Govern Flow Motion
36 relationship is given here without any comment for the (1) is irrotational; (2) satisfies the continuity equation; (3) is
simple purpose of recognition in literature: such that a/az = 0 for z = -d.
a(x,y,z) =1 3.7 Derive the continuity equation and V2 for an incom-
a(x 0 ,y0 ,z0 ) pressible liquid in spherical polar coordinates {r,O,CI>) by
where x 0 , y 0 and z 0 are the coordinates of the considered considering a small volume bounded by the surface:
particle at time t 0 and x, y, z at time t. r, r + dr; 0, 0 +dO; Cl>, Cl> +del>.
~ (av _au) dx
2 ax ay
To each of these velocity components corresponds a
component of acceleration, which multiplied by p, yields Figure 4-1 Periodic gravity wave in deep water.
a component of inertia force. The velocity at a given point changes in
Two types of inertia forces may be distinguished, direction only.
depending on the type of acceleration or elementary motion
considered. These are:
Alternatively, if the velocity decreases, which corresponds
1. Local acceleration -corresponding to a variation of
to a negative local acceleration, the inertia of the mass
the velocity of translation or the derivative of velocity
of fluid in motion tends to maintain the original velocity.
with respect to time. This type of phenomenon is observed in pipes or tunnels,
2. Convective acceleration-corresponding to a varia- where a fluid stops or starts or balances because of a gate
tion of velocity of deformation and rotation or movement. Local inertia has to be taken into account in
derivative of velocity with respect to space. hydraulic engineering applications such as surge tanks,
The physical meaning of these accelerations and the water hammer, and locks.
corresponding inertia forces is first examined; then their
mathematical expression is demonstrated. Chapter 5 4-2.1.2 In the second case, the velocity maintains the
deals with the applied forces F which have to be equated same magnitude, but changes its direction. In this case
to these inertia forces to obtain the momentum equation. the inertia force is due to the centrifugal acceleration.
For example, in a periodic gravity wave in infinite depth,
4-2 Local Acceleration the magnitude of the velocity at a given point is a constant
but its direction revolves continuously at all points (Fig. 4-1 ).
Local acceleration characterizes any unsteady motion,
i.e., motion where the velocity at a given point changes with 4-2.1.3 In the third case, the velocity changes at a given
respect to time. Local acceleration results from a change
point both in magnitude and direction. Some examples
in the translatory motion of a fluid particle imposed by
of this case are turbulent flow (this important case is
external forces F.
fully developed in Chapters 7 and 8); alternate vortices;
4-2.1 Examples ofF/ow with Local displacement caused by a ship in motion; bore and tide
Inertia in an estuary; and periodic gravity waves in shallow water.
Local acceleration occurs in the following cases.
4-2.2 Mathematical Expression of Local
Inertia
4-2.1.1 In the first case, the velocity stays in the same
direction along a straight line and changes in magnitude. The mathematical expression of the inertia forces
If the velocity increases at a given point, which involves a caused by a local acceleration is given by the change in the
positive local acceleration, the inertia of the mass of fluid velocity of the translatory motion with respect to time only.
in motion tends to slow it down. The corresponding inertia force is equal to p(av;at) of
Chapter 4: Inertia Forces
which the components along the three axes are, respectively: The expressions aujax, ovjoy, and owjaz are taken at a 39
p(oujot), p(ovjot), and p(owjot). The derivatives with respect given time, as seen in section 2-3.1. The corresponding
to space are not taken into account. acceleration is
_t1_ [au dx] = ou dx
dt ax ax dt
4-3 Convective Acceleration
Two similar expressions result for w and v. If u = dxjdt,
Convective acceleration characterizes any nonuniform v = dyjdt, and w = dzjdt, are substituted in these expres-
flow, i.e., when the velocity at a given time changes with sions and the result is multiplied by the density, the inertia
respect to distance. It is sometimes called field acceleration. forces are obtained. They are
Convective acceleration results from any linear or angular
au 1 a(u 2 )
deformation, or from a change in the rotation of fluid pu-=-p--
particles, imposed by external forces F. ax 2 ox
ov 1 a(v 2 )
4-3.1 The Case of Linear Deformation pv- = -p--
oy 2 ay
4-3.1.1 In a convergent pipe, it has been seen that the ow 1 o(w 2 )
velocity of a fluid particle, although constant with time at a pw-=-p--
az 2 oz
fixed location, tends to increase along the converging
streamlines. The velocity of the fluid particle increases It should be noticed that the last group of expressions
with respect to space. This is a positive convective accelera- may be written as (ajox)(pu 2 j2). This shows that the inertia
tion. The fluid tends to resist this acceleration by convective force is equal to the derivative of the kinetic energy with
inertia. respect to space along the three direction axes OX, OY,
In a divergent conduit, the velocity decreases and the and OZ, respectively.
fluid tends to continue its motion with the same velocity
4-3.2 The Case of Shear Deformation
because of its inertia. The applied forces cause a negative
convective acceleration. 4-3.2.1 In a bend, where the fluid particles are angularly
Expansion or contraction of a compressible fluid is the deformed, the fluid paths are curved and because of its
sum oflinear deformations and also results in corresponding inertia, the fluid tends to continue along a straight line.
inertia forces. This causes a centrifugal force proportional to the change
of direction which is imposed by the applied forces.
It is possible for the velocity of a fluid particle to keep
4-3.1.2 It has been seen that the linear deformation
the same magnitude along its path, but with a change in
velocity components are those given in Equation 4-1.
direction. This is the case of free vortex motion.
~~dx}
4-3.2.2 It has been seen that the velocity components
Two-dimensional of angular deformation for a two-dimensional motion are
motion
-av d y Three-dimensional
(4-1) ~(au+ av) dy
oy motion 2 ay ax
ow dz 1 (au av)
az 2 ay +ox dx
Part 1 : Establishing the Basic Equations that Govern Flow Motion
40 Hence, as in the previous case, using the substitutions are the velocities of the components of rotation in a two-
u = dx/dt and v = dy/dt, the corresponding inertia forces dimensional motion, the corresponding inertia forces
become: obtained are
1
2 pv
( au av)
oy + ox
- ~ pv(av - au)
2 ax ay
~ pu(ov _ ou)
2 ax ay
4-3.3 The Case of a Change of Rotation It has been shown that it is possible to assume that the
motion is irrotational when friction effects are negligible.
4-3.3.1 In the entrance to a pipe (Fig. 4-2), because of the
It is evident that the same conditions lead to neglect of
change in friction forces, there is a variation of rotation
rotational inertia forces.
of the fluid particles. Hence there are inertia forces cor-
responding to the natural resistance of the fluid to change
its rotational motion. In a uniform pipe, the rotation of
particles exists but there is no change in rotational magni- 4-4 General Mathematical Expressions
tude and the corresponding acceleration is zero. of Inertia Forces
4-3.3.2 As in the two previous cases, since 4-4.1 Local and Convective Acceleration
- ~2 (av- au) dy
ax ay
In the general case both local acceleration and convective
acceleration occur at the same time. A simple example is
when a fluid oscillates in a nonuniform curved pipe. Hence,
~ ( av _au) dx in the general case, V and its components u, v, and w are
2 ax ay
functions of both time and space coordinates. For example,
u(x,y,z,t). The total differential of u is
(auat +
ou ou ou)
u-+v-+w-
expression, valid along the 0 X axis: 4I
P ax oy oz ( ou ov ow)
[au
ot u-+v-+w-
p-+ ax ax ax
ov ov ov)
u-+v-+w-
Pev
ot + ax ay oz (4-2)
+ v(ou _ ov) + w(ou _ ow)]
oy ax az ax
(ow ow ow ow)
u-+v-+w- But
Par + ax oy oz aw = -1 -
au + vav- + w- a-
a (u 2 + v2 + w2 ) = - V2
u-
'--,-1
Local Convective
OX OX ox 2 ox ox 2
acceleration acceleration When the coefficients of the rotational vector
terms terms
2'1 =(au _ ow)
4-4.2 Elementary Acceleration oz ox
Components
2( = (ov _ ou)
Following a procedure similar to that used in the study ax oy
of the elementary motions of fluid particles (Section 2-6.2), are introduced, the following expression for the inertia
that is, adding and subtracting !pv(ovjox) and !pw(owjox) forces along the 0 X axis results
to the first line above, gives Equation 4-3, which emphasize
the previous physical considerations.
Similar forms can be obtained for they and z components
p[~~ + :X (~2 ) + 2(1]W - (v)J
of the forces. Similarly, it may be found that the inertia forces along the
OYand OZ axes are
4-4.3 Separation of Rotational Terms
It is often useful to transform the acceleration terms p - [ov a (V
at + -oy -2
2
) + 2((u - ~w) J
to a form which emphasizes both the kinetic energy
terms and the rotational terms. Adding and subtracting
p[v(ovjox) w(owjox)]
+ to the first line, gives the following
p - [ow + -a (V
at az 2
2
- ) + 2( ~v - IJU) . J
[au ou
p- + u- + (4-3)
at ox
'-v--1 '-v--1
Local Acceleration Acceleration Acceleration in
acceleration in linear in angular rotation
resulting in deformation deformation
a change in
translatory Convective acceleration terms
motion
Part 1 : Establishing the Basic Equations that Govern Flow Motion
These three expressions may be written more concisely outlet control, and turbined discharge; and (4) emptying
in vector notation as shown in Equation 4-4. of a basin by a small valve. In all these cases, the flow is
considered as a succession of steady motions and calculated
av
(
pat+ grad 2
V2
+ curl Vx V). (4-4) as such without taking account of local inertia.
~ '-.,-' '-.-'
4-5.1.2 As an example of unsteady motion analyzed
Local Kinetic energy Rotational
acceleration term term as a succession of steady motions, the variation of level
in a basin is studied.
Convective acceleration Consider the emptying of a rectangular basin of hori-
It has to be noticed that the convective inertia term
zontal cross-sectional area A. The volume of water above
a small hole of area S is A times z. The variation of z as a
p grad ~z =p[i :x (~2) + j ;y (~2) + k ;z (~2)J function of time is given by the differential equation
Adz = Q dt where Q = CdSV, Cd being the coefficient
of discharge and V being given by the formula of Torricelli:
is, in fact, the derivative with respect to space of the kinetic V = (2gz) 112 . The formula of Torricelli is well known in
energy, pV 2 /2, of the particle. elementary hydraulics, but it is important to point out
herewith that this formula is only valid when the local
4-5 On Some Approximations inertia is negligible.
Introducing the value of Q = CdS(2gz) 1 12 in the above
4-5.1 Cases Where the Local equation and integrating gives the total time required to
Acceleration Is Neglected empty such a basin, (z 0 is the initial depth)
4-5.1.1 A rigorous steady motion never exists. There
is always a beginning and an end. However, many motions - 2A 1/2
T- CdS(2g)ll2 (zo)
in hydraulics are actually very close to being steady during
a given interval of time. In this case, since V does not vary If S were large and A small, it would be necessary to take
with time, the corresponding inertia term av jot is zero. account of the local inertia to calculate T.
(The very important case of turbulent motion will be Another similar example, previously cited, is that of the
studied in Chapters 7 and 8.) variation of level in a reservoir of horizontal section S(z)
However, there are many unsteady motions in hydraulic because of the variation of the upstream river flux. The
engineering in which the local acceleration and the cor- correspondi"ng calculation of the economical height of the
responding inertia terms are neglected. dam, the number of turbines and the spillway capacity
This occurs when the velocities are slow and their are deduced from Equation 4-5.
variations with time are very slow. For instance, in the
case of a periodic motion in which the period Tis very long:
4-5.2 Cases Where the Convective
oVjot ~ V/T Hence p(oVjot) would be negligible m
Acceleration Is Neglected or
comparison with other forces.
Approximated
Some particular cases where this approximation is
valid are (1) flow in a porous medium: variation of the 4-5.2.1 The local inertia term is proportional to the
ground water table with respect to time; (2) flood wave velocity V and thus, it is a linear term. The convective
in a river; (3) variation oflevel in a reservoir because of the acceleration terms are quadratic and are proportional to
variation of the upstream river flux, the spillway and bottom V 2 (or a product u2 , v2 , w2 , uv, uw, vw). Since the convective
Chapter 4: Inertia Forces
43
Q(t) Llt = S(z) Llz + Q1 Llt + Q.(z) Llt + f(S) Llt (4-5)
'-v--' ~ '-v--' ~ ~
Upstream Change of Turbined Volume Loss by
influx of level in volume over the evaporation
the river reservoir spillway
acceleration introduces a quadratic term, the general 4-5.2.4 Sometimes only some terms of the convective
equation of momentum is nonlinear. acceleration may be neglected. The case of a two-dimen-
It is well known that it is easy to mathematically solve sional boundary layer on a flat plate is given here as an
many linear differential systems of equations. But it is example (Fig. 4-3). This example is particularly helpful
often difficult to solve a nonlinear system. This is the chief in understanding how the mathematical simplifications
cause of difficulty in fluid mechanics. For this reason, it is may be based on physical considerations. Hypothesis:
helpful to know when it is possible to neglect this quadratic u is large in comparison with v; the derivatives with respect
term. to y are large compared to the derivatives with respect
When V tends to zero, a quadratic term proportional to to x. The continuity equation for two-dimensional motion
V 2 tends to zero more rapidly than a linear term pro- shows also that oujox and ovjoy are of the same order.
portional to V. Hence, in practice, when V is small, V 2 is (Section 3-1.2). Hence, the 0 Y components of the convective
negligible and the convective inertia term is negligible in inertia terms are negligible because
comparison with the other terms expressing the local
inertia and applied forces. Such motions are called "slow
au au ov ov
u-+v-~u-+v-
motions." ox ay ax ay
Some examples of such motion are: (1) periodic gravity Similar approximations are made to analyze the develop-
wave theory (first order of approximation); (2) flow in a ment of a jet, and in the nonlinear long wave theory
porous medium, which obeys the linear law of Darcy (Chapter 18).
(such motion is defined only by an equality of applied forces
since the local acceleration is negligible); and (3) motion
of a small sphere in a viscous fluid (Stokes' formula). Figure 4-3 Introduction of simplifying
assumptions in the theory of
4-5.2.2 Sometimes a partial effect of the convective development of a boundary layer.
acceleration is taken into account by the use of an approxi-
mate solution given by a number of terms of a series.
(Example: gravity wave theories to the second order,
third order, etc. of approximation; laminar boundary layer v
theory; etc.)
45
DIRECTION OF THE
~ EARTH'S ROT/lrtON
EARTH'S EARTH's
AXIS AXIS
Figure 4-5 The Coriolis effect in the southern Figure 4-6 The Coriolis effect in the eastward and
hemisphere. westward motion in the northern
hemisphere.
N
alone, the fluid particle A has a velocity r(J) toward the
east. ((!) is the angular rotation of the earth.) The velocity
components relative to the earth are U E = r(d()jdt) toward
the east, and (U 8 sin <I>- W cos <I>)= drjdt along BA.
Therefore, the resultant velocity has an easterly component
r[(J) + (d()jdt)] and a radial component drjdt.
us
4-6.2.3 Consider a vector defined by the two components
X = R cos a and Y = R sin a. The derivatives of this vector
with respect to time are given by the derivatives of its
components, i.e.,
dX dR . da
- = - cos a - R Sill a -
dt dt dt
N dY dR . da
- = - Sill a + R cos a --
dt dt dt
The derivative of this vector is thus given by a sum. The
B first part, in the direction of the vector, has the magnitude
dRjdt. The second part has a magnitude R(dajdt).
Since
N B
z 47
r~
dt
d2 r
X
dt2
M Figure 4-10 Components of acceleration.
dr s
dt in the easterly direction, and a radial component equal to
Figure 4-9 Components of the derivative of a
vector r. r(w + ~~y
The total acceleration is composed of these four com-
from Fig. 4-9 that the component r(dejdt), perpendicular to ponents, as shown on Fig. 4-10.
the vector (X, Y) is actually to the east of this vector. If the radial component r[w + (dejdt)] 2 is written as
r[w 2 + 2w(d8jdt) + (dejdt) 2 ], further observations can be
4-6.2.4 The previous results will be applied to both the made.
radial component, dr/dt, and the easterly component, 1. The terms which are independent from w
r[w + (dejdt)], of the velocity vector. The acceleration
or time rate of increase of drjdt has a component d2 r/dt 2
in the same radial direction which is due to centrifugal
~;~ 2e:)(~~) + r ~:~ and -r(~~y
acceleration, and an easterly component perpendicular are the acceleration of the particle relative to the earth.
to BA which is equal to 2. The term rw 2 , directed toward B, is the centrifugal
acceleration.
dr
dt
(w +de)
dt
3. The two terms 2w(drjdt) and 2wr(dejdt) are per-
pendicular to and to the left of the velocity com-
ponents
Similarly, the acceleration or time rate of increase of the
easterly component {r[w + (dejdt)]} has a component, dr . --" de --"
dt = Us Sill 'I' - W COS 'I' UE and r-=
dt
respectively and are the geostrophic acceleration
terms.
Part I : Establishing the Basic Equations that Govern Flow Motion
4-6.2.5 The corresponding inertial components are geostrophic terms in this system of coordinates are:
attained by multiplying these accelerations by the density - nv along the 0 X axis and + Qu along the 0 Y axis.
p. Therefore the corresponding geostrophic inertial forces Adding the local and convective forces, with the vertical
are component w neglected, the total inertial forces are
2pw(U s sin <I> - W cos <I>)
au + u-
p( -
au - nv )
au + v-
directed horizontally toward the east, and at ax ay
4.3 In the case of a progressive acoustic wave in a pipe, 5. When a constant discharge q0 = 1 fe /sec (28,317 49
such that em 3 /sec) is poured into the tank, establish the function
IJ(t) ('1o = 20ft, 609 em) and determine '1 when t ~ oo.
. 2n
u = a sm L (x - Ct)
The discharge coefficient Cn will be assumed to be unity.
calculate the ratio of convective inertia to local inertia. 4.5 Determine the convective inertia terms which can be
neglected in a jet.
4.4 Consider a basin such as shown in Fig. 4-11, where the
particle velocity at the orifice is V = (2g1J) 112 , z = '1 is the 4.6 The influx of discharge into a reservoir is defined by
height of the free surface above the orifice. The horizontal the equation
area of the basin being A = 4 ftl, (3716 cm 2 ), determine:
Q(t) = 10,000 (1.5- sin kt) ft3/sec
1. The variation of the free surface with respect to time, = [283 (1.5- sin kt) m 3 jsec]
i.e., the function l](t). At timet = 0, '1o = 20ft (609 em).
where k = 2n/T and Tis a period of 1 year. The horizontal
The local inertia will be neglected for this calculation.
area of the reservoir is defined by
2. The local inertia at point x = 3 ft (91.44 em) z = 0,
i.e., in the converging section. A(z) = 10,000z2 ftl (929z 2 m 2 )
3. The convective inertia at the same point.
The top of the spillway for flood discharge is located at an
4. Is the neglect of the local inertia a valid assumption?
elevation z 1 = 100 ft (30.48 m) and has a discharge capacity
Explain. The friction will be neglected.
Qs = C1[z - z 1] [2g(z - z 1 )] 112
where the coefficient of discharge C = 0.5, and the length
Figure 4-11 of the spillway 1 = 100 ft (30.48 m). The turbined discharge
is constant and equal to 7000 ft 3 /sec (198m 3 /sec).
Determine the variation of the level of the free surface in
the reservoir as a function of time and the maximum
discharge over the spillway for each year following time
t = 0. The maximum possible discharge over the spillways
will also be determined. At time t = 0, one will take the
free surface elevation at z = 30ft (9.144 m).
d o o UE o Us o
-=-+
dt dt
W-+---+--
or r sin <I> oO r o<l>
sr
Chapter 5 5-l Internal and External Forces
The applied forces on an elementary mass of fluid
consist of internal forces and external forces.
L.
-ojoz(pgz). In vector form, they are -grad(pgz), since
X= -o/ox(pgz) = o and Y = -o/oy(pgz) = o.
The gravity force is often neglected in gas dynamics
except, for example, in meteorology or in the calculation
for chimneys and ventilation openings, where the pheno-
mena are influenced by the variations of gravity forces
due to density changes.
Chapter 5: Applied Forces
the pressure is seen to be the same in all directions. A Hence, the rate of change of pressure force per unit of 53
similar demonstration is possible in three dimensions. volume is given by the three components -ap;ax, -apjay,
However, it is evident that the gradient of pressure force and - apjaz, which can be written vectorially: -grad (p).
(which is a vector) changes with direction. In the same way,
the force caused by pressure against an area (which is a 5-3.3 Fluid Motion and Gradient of
vector) changes direction as the normal to the considered Pressure
area changes direction. It is interesting to note that the motion of the fluid
particle does not depend upon the absolute value of p, but
5-3.2 Rate of Pressure Force Per Unit only upon the gradient of p. Consider the motion in a
of Volume tunnel. The motion depends upon the difference in pressure
Consider an elementary fluid particle (dx dy dz) (Fig. 5-2). levels between the upstream and the downstream sections.
The pressure force due to the external adjacent fluid Therefore, it is possible to operate a scale model experiment
particle acting against the side ABCD is p(area ABCD) = at any convenient arbitrary absolute pressure, provided
p dy dz. The pressure force against the other side acts in the pressure gradient is kept in similitude. However, if the
the opposite direction and may be written: pressure level drops below a critical value, cavitation occurs
and, for similitude, the scale model must be operated
- (P + ~~ dx) (area EFGH) = - (P + ~~ dx) dy dz under partial vacuum (see Appendix B).
5-3.4 Pressure and Gravity
Hence, the difference of pressure forces acting in opposite The total force due to the pressure force and gravity force
directions is per unit volume is
ap ) ap grad p + grad pgz = grad (p + pgz)
p dy dz - ( p + ax dx dy dz = - ax dx dy dz
The sum of these two linear quantities (p + pgz) is a
Similarly, the pressure force differences acting in the constant in hydrostatics since p - Pa = - pgz where Pais a
constant external pressure (atmospheric). This property is
OY and OZ directions are -(apjay) dx dy dz and
also verified in a cross section of a uniform flow as in a
-(ap;az) dx dy dz.
channel or in a pipe, or more generally when the curvature
of the paths is negligible or the motion is very slow (see
Section 10-2.1.1). Hence the sum (p + pgz) may often be
Figure 5-2 Difference in pressure forces in a unit conveniently replaced by the single term p*: p* = p + pgz.
of volume. In hydrostatics p* =constant. Whereas pjpg is known as
the pressure head, p* / pg is called the piezometric head.
p
5-4 Viscous Forces
5-4.1 Mathematical Expressionfor the
Viscous Forces
Shear stresses are present because of fluid viscosity and
are caused by the transfer of molecular momentum.
The friction force r is assumed to be proportional to the
Part 1 : Establishing the Basic Equations that Govern Flow Motion
G H
54 More generally, for a three-dimensional incompressible
fluid, it is possible to demonstrate that the friction force
u+~dy components per unit of volume are:
oy
y
c 02U 02U 02U)
L.
D
J1 ( ox2 + oy2 + oz2 = J1'V2u
advantage of so expressing the surface forces lies in its Hence, the external forces may be defined by a tensor of 55
applicability to any kind of motion, e.g., perfect, viscous or rank two:
turbulent, compressible or incompressible. However, the
values of these surface forces are expressed differently when
their physical nature is taken into account.
SHEARING
FORCE
/
/
T XZ /
I OT xy
NORMAL r,Y + ~ dx
FORCE
I
I
Figure 5-4
I I
Notation for surface forces. I
lo
}-':--- -
/
/
/
A E
y OT xz
Txz + ~dx
Part 1 : Establishing the Basic Equations that Govern Flow Motion
ox OY oz
On the side of area { ABCD EFGH DHGC AEFB AEHD BFGC
(see Fig. 5-4) dydz dy dz dxdz dx dz dx dy dx dy
ox ~XX
( ~XX+ a~xx dx )
a;: Tyx (
Tyx
oryx )
+ aJdy '=x
(
r zx + ---a;
OTzx
dz )
oz Lxz (
r xz + ---a;
OTxz
dx ) Tyz ( Tyz oryz )
+ aJdy ~zz
(
~ ==
a~zz dz )
+ ---a;
The total sum of the torque caused by the shearing Similarly, it may be shown that ryz = rzy and rxz = rzx
stresses is Hence, the nine components of the tensor of external force
are reduced to the six so-called components of Lame.
rxy(dy dz) dx - ryx(dz dx) dy
This torque is equal to the mass times the square of the
radius of gyration (dRf, times the square of the angular
velocity (w 2 ), which may be written p dx dy dz(dR) 2 w 2
Since dR is infinitesimally small, having the same order
as dx, dy, and dz, (dR) 2 is of the second order and the speed 5-5.3 Value of the Lame Components in
of gyration becomes infinite, which is physically impossible. Some Paracular Cases
Hence, the total torque must be zero. This condition is 5-5.3.1 In the case of a perfect fluid, the shearing stresses
possible only when rxy = ryx are zero and the normal forces become simply the pressure
forces:
(Jxx = (JYY = (Jzz = - P
Figure 5-5 Torque about the point A. rxy = !yz = !zx = 0
+ 2f.1-
au
== -p
(J XX
ax
Similar expressions exist for CJ YY and CJ ==.
Chapter 5: Applied Forces
The shearing stresses r are functions of the coefficients The work of all the forces in the 0 X direction is: 57
of angular deformation:
oa xx ) dydz ( u+ oxdx
(Jxxdydzudt- ( (Jxx+~dx au ) dt
+ 11[2(au)
ox + 2(av)
oy + 2(aw)
5-5.3.3 In the case of a viscous compressible fluid, the 2 2 2
shearing stresses are the same as in the above case, but the <l> = A(div V) 2 oz
normal forces have to take into account the change of
volume of the fluid particle. It may be seen that: + (aw + av) 2 + (au + aw) 2 + (av + au) 2 ]
oy az az ax ax ay
au av ow) au
axx = - P + A ( ax + ay + az + 211 ax A(div V) 2 is equal to zero in an incompressible fluid.
This function can be used, for example, in the calculation
of head loss of a viscous flow in a pipe and the damping of
Two similar relationships are easily deduced for ayy and
gravity waves.
a zz- A is a second coefficient of viscosity for a gas. From the
kinetic theory of gases, it may be shown that for a mono-
atomic gas 3A + 211 = 0. In practice this relationship is
considered accurate enough for any kind of gas. PROBLEMS
ap*
ax
8p*
ay
dw 8p*
pdt az
Written in vector notation, these become
dV
p dt + grad p* = 0
6o expanded (see Section 4-4.1), the momentum equation Two similar equations give the value of opfoy 0 and opfoz 0 by
takes the form along the OX axis given by Equation 6-1. permutation of x 0 , y 0 , z 0 , which are the coordinates of a
particle at time t = t 0 . These are called the equations of
Inertia forces Applied forces Lagrange.
(-
p
ou) =
ou + w-
ou + v-
au + u- a
--(p + pgz) (6-1) 6-1.2.1 If the friction forces are introduced in the Eulerian
at ax oy az ax equations, the Navier-Stokes equations are obtained (see
Section 5-4.1), as shown in Equation 6-2.
Two similar equations may be written in the OY and OZ The Navier-Stokes equations are the basis of most
directions. These are called the equations of Euler. problems in fluid mechanics dealing with liquid. They are
Such a system of equations associated with the continuity second-order differential equations because of the friction
relationship oufox + ovfoy + owfoz = 0 forms the basis terms, and nonlinear because of the convective inertia
of the largest part of the hydrodynamics dealing with a terms.
perfect incompressible fluid. These equations are mathe-
matically of the first order but are nonlinear (more specif- 6-1.2.2 These Navier-Stokes equations are written in a
ically quadratic) because of the convective inertia terms. very concise manner with the aid of tensorial notation.
This quadratic term is the cause of most mathematical Although a knowledge of tensorial calculus is not required,
difficulties encountered in hydrodynamics. it is given here as a guide to further reading on the subject.
Use is made of two subscripts, i and j, which indicate
6-1.1.2 It has been explained in Chapter 1 that it is when an operation is to be systematically repeated and
possible to study hydrodynamic problems either in Eulerian which component of a vector quantity (such as V) is being
coordinates or in Lagrangian coordinates. It is recalled considered. When an index is repeated in a term, the
that the Lagrangian method consists of following particles considered quantity has to be summed over the possible
along their paths instead of dealing with particles at a components. For example, the continuity equation
given point. This method is used, for example, in some oufox + ovfoy + owfoz = 0 is tensorially written: ou;/ox;
studies of periodic gravity waves over a horizontal bottom. = 0, since the subscript i indicates that the quantity (here
The corresponding equations are given here only for the V) has to be summed over the three components OX,
purpose of recognition in the literature and will not be OY, OZ.
developed. The three previous Navier-Stokes equations, may be
If X, Y, Z are the volume or body forces, i.e., gravity, written simply as:
the Lagrangian equation along the OX axis is written:
61
Inertia forces Applied forces
op
ox
ov ov ov ov) op
(
p~+u~+v~+w~ = (6-2)
ot ox oy oz oy
p( 8t
OW OW
+ u OX + v oy + w
OW OW)
oz = -
o(p +
oz
pgz)
+ f.1
(oox
2w
2 +
o2 w
oy 2 +
o 2 w)
oz 2
6-1.2.3 These Navier-Stokes equations are often written Since the derivative of the sum in parentheses is zero in all
in another way in order to emphasize the role of the directions, one obtains
rotational component of motion. It is sufficient in this case
to use the expression of the inertia force demonstrated
v2
pT + p + pgz = constant
in Chapter 4, which yields (see Section 4-4.3) Equation 6-3.
which is the well-known Bernoulli equation, fully developed
6-1.2.4 The three components of Equation 6-3 are more in Chapter 10.
concisely written in the vector form of Equation 6-4, which
may be transformed as 6-1.3 The General Form of the
Momentum Equation
grad ( pv- + p + pgz)
2
= -
ov
p - - p(curl V) x V + f.1V 2 V It has been shown that the applied forces may be ex-
2 ot pressed independently of their physical nature with the help
of the tensor of rank two:
6-1.2.5 In the case of a steady (oV jot = 0) irrotational
flow (curl V = 0) of a perfect fluid (f.l = 0), the above
equation gives at once:
v2
grad ( pT+ p + pgz )=o The main advantage of such a notation is that it is valid
for any kind of fluid-perfect or real-and any kind of
Part 1 : Establishing the Basic Equations that Govern Flow Motion
Caused by Caused by
variation rotation
of kinetic
energy
p[ -ou
ot
(V
+ -o -
ox 2
2
) + 2(w1J J
- v() = - o(p + pgz) +
OX
n2
p,v U
p[ -ov
ot
+ -o -
oy
(V2 2
) + 2(u( - w~) J= - o(p +
oy
pgz) nz
+ p,v v (6-3)
p [ -ow
ot
(V
+ -o -
oz 2
2
) + 2(v~ J
- U1J) = - o(p
+ pgz)
OZ
n2
+ p,v W
Caused by Caused by
variation rotation
of kinetic
energy
dv
p dt (6-5)
motion-laminar or turbulent. It will be shown that if approach. The motion is averaged and the friction term
in the momentum equation the real values u, v, w, and p J1'V 2 V is replaced by an empirical functional relationship
are replaced by the average values ii, v, w, and p in a tur- proportional to V in the case of flow trough porous
bulent flow, the surface forces (J and r include additional medium, or to V 2 in the case of fully turbulent motion.
components due to the turbulent fluctuations. These two points are analyzed in Chapters 8 and 9.
Hence, the advantage of using the notations (J and r
exists in expressing general equations which are independent
of the nature of the flow. Equating the inertia forces to the
applied forces expressed in the manner shown in Chapter 5 6-2 Exact Integration vs Numerical
yields Equation 6-5. Solutions
In practice, if OZ is vertical upward, X = 0, Y = 0,
z = - pg = -(ojoz)(pgz). 6-2.1 An Example of an Exact Solution
of Navier-Stokes Equations. Flow on a
6.1-4 Synthesis of the Most Usual Sloped Plane
Approximations
6-2.1.1 It is to be expected that a general solution of the
Tables 6-2 and 6-3 recall the physical meaning of system of differential equations given by the continuity
different terms and possible approximations accepted and momentum principle does not exist. However, some
in the studies of flow motions, which may be investigated exact solutions can be obtained if the boundary conditions
in the following. Complex disordered and random motions, are simple. Examples where exact solutions may be obtained
even though also obeying the Navier-Stokes equation, include flow between parallel plates (i.e., the Couette
cannot be analyzed on a purely Newtonian deterministic flow, the Poiseuille flow), flow due to a rotating disk,
Part I : Establishing the Basic Equations that Govern Flow Motion
64 Table 6-2
[av
Par + grad(~2 ) + (curl V) x VJ - grad(p + pgz) + J.i'V 2 V
Mathematical First-order Nonlinear (quadratic) term First -order Constant term Second-order
characteristics linear linear linear
term term term
=0 =0
for slow motion
uniform unsteady flow over an infinite flat plate (see The Navier~Stokes equations are reduced to:
Section 13-2.3.3), etc.
pg sin a+ J!G:~) = 0
6-2.1.2 The very simple example of a two-dimensional
steady uniform flow on an inclined plane of infinite dimen-
ap
0 = - - - pg cos a
sions is given here as an example (Fig. 6-1); the Navier~ ay
Stokes equation given in Section 6-1.2.1 may be simplified The second equation yields
in the following manner: p = Pa - pgy cos a,
Since the motion is steady, aujat = 0 and avjat = 0.
Since the motion is two-dimensional, w = 0, and all where p. is the atmospheric pressure. Hence the lines of
derivatives with respect to z are zero. Since the motion is equal pressure are parallel to the OX axis.
uniform and parallel to the axis OX, v and all its derivatives The boundary conditions are u = 0 for y = - d on the
are zero. All derivatives with respect to x are also zero. plane, and dujdy = 0 for y = 0 at the free surface. Taking
The components of the gravitational force are X = pg sin a into account these boundary conditions, the integration of
and Y = - pg cos a. Since the flow is uniform, v = 0, and a2 u = - -sma
-
pg .
the continuity equation is reduced to aujax = 0. ayz Jl
Chapter 6: Forms of the Momentum Equation: Equations of Euler and Navier-Stokes
Table 6-3 6s
Convective
Local inertia inertia Friction Equations Some applications
av =0
+pat+ p(curlV) x V
and = g sin IX d 3
_ g sin IX (d 2 -y2)
U---
3v
2v
The loss of energy per unit length may be given by the
which is the equation of a parabola. dissipation function <1>, which in this case is J.l(oujoy) 2
Part 1 : Establishing the Basic Equations that Govern Flow Motion
------
Instabilities initially occur either within the fluid or
at a boundary. In the first case the phenomenon is called Figure 6-4 Instability of a laminar flow because of
"free turbulence." The lumps of turbulence come initially a roughness.
from the zone where the gradient of kinetic energy is a
maximum, as for example from the boundary of a submerged
jet. Hydraulic jumps, breaking waves on a beach, and ~
1 111171/lj)lllllll
whitecaps generated at sea under wind action, are also
cases of free turbulence. Initial instabilities are most often
caused by roughness of a fixed boundary. Indeed, any -------------==
------~
___.....--- llA u~
roughness causes a local increase of velocity which con- ..,.1..,1-,7-,l-1,..-,1;T''// II! II I l l /
sequently produces a local scrong gradient of kinetic
energy, causing an instability (Fig. 6-4). This instability
also arises between two fluids of different density. For
example, the wind blowing on a liquid causes ripples.
~
I I I I I I I f/>.1 I I I I I II
Chapter 6: Forms of the Momentum Equation: Equations of Euler and Navier-Stokes
Reynolds numbers, laminar flow is unstable if the distur- 6.4 Calculate the two-dimensional velocity distribution 6g
bance is large enough. It is possible, with many precautions, u(z) between two parallel horizontal planes between which
to obtain a laminar flow in a very smooth pipe up to a there are two layers of fluid of thickness e 1 and e 2 , viscosity
Reynolds number of 40,000, although under normal J1 1 and Jlz, and density p 1 and p 2 (p 1 > p 2 ), respectively.
conditions the critical value of Reynolds number for a pipe One plane is fixed and the upper plane moves at constant
is 2,000. A disturbance superimposed on the primary velocity U.
motion causes, as previously seen, a large local increase
in the convective inertia forces.
6.5 Consider a two-dimensional flow between two fixed
parallel horizontal planes separated by a distance 2h.
70 6.7 The following dimensionless quantities are defined: 6.9 Demonstrate that the Navier-Stokes equation can
still be written:
z
x* = -
X
y* =I z* =-
L L L p d~ - ll'~jl~ = p[~ ou + 11 ou + 'ou]
dt ax oy oz
t V*(u*,v *,w *) -_ V(u,v,w)
t* = -
T
u Obtain the two other equations by circular permutation.
72 stability of the solution, that is to determine whether a sum of the mean pressure p and a fluctuation term p' such
small disturbance will increase or be damped out by friction. that p = p + p' where
However, a fully deterministic approach is no longer
possible in the case of turbulent motion, because of the p= T
1 fT p dt and p' =
1
T
fT p' dt = 0
0 0
random nature of these turbulent fluctuations.
On the other hand, in engineering practice, it is not Hence, turbulent motion may be considered as the super-
always necessary to know the exact fine structure of the position of a mean motion and a fluctuating and disorderly
flow. Only the average values and the overall and statistical motion, random in nature, which obeys statistical laws.
effects of turbulent fluctuations have to be studied.
7-1.4 Steady and Unsteady Mean
7-1.3 Definitions of the Mean Values in a
Turbulent Flow Turbulent Flows
In a turbulent motion, as in the case of a viscous flow, The mean value is defined for intervals of time T, which
velocity and pressure have to be known as functions of the are large compared to the time scale of turbulent fluctua-
space coordinates and time. tions but small compared to the time scale of the mean
The instantaneous velocity V at a fixed point is the motion.
vectorial sum of the mean velocity V with respect to time If, for example, one considers the oscillation of water in a
(referring to the basic primary movement) and .the fluctua- tunnel between a surge tank and a reservoir, the instan-
tion velocity V' which varies rapidly with time both in taneous velocity at a fixed point may vary quickly because
intensity and direction. This can be expressed by the of the turbulence. The average velocity defined for a
relationships V = V + V' where, by definition, relatively short interval of time varies also with respect to
time, but its change is slow.
V = _!_ fT V dt The real motion is always unsteady because of turbulence
T o and in this case, the mean motion is also unsteady (Fig. 7-1).
and In the following discussion, a motion is called unsteady
only if the mean value of the velocity varies. The interval
of time, which permits a realistic definition of the mean
V' = _!_ fT V' dt = 0
motion, is relative to the frequency of turbulent fluctuations.
T o
It varies with the phenomenon to be studied. For example,
where T is a time interval. Similarly, the instantaneous it is long for the meteorologist who deals with atmospheric
components of velocity are defined as follows: motion, and it is short for the aerodynamist who deals
with the turbulence effects in the boundary layer along a
Real Mean Fluctuation wing.
velocity velocity velocity
7-1.5 Mean Forces
u ii + u'
v i5 + v' Since the real value of the inertia forces is always equal to
w w + w' the sum of the real values of the applied forces in any kind
of motion (laminar or turbulent), the mean value of the
inertia forces with respect to time is equal to the mean
and u' = v' = w' = 0. value of the applied forces with respect to time. This may
Similarly, the instantaneous pressure p IS the scalar be expressed as shown in Equation 7-1. Since
Chapter 7: Turbulence: Mean and Fluctuating Components of Motion
v v 73
Figure 7-1
The steadiness of a turbulent flow is
defined by the mean velocity only.
Vaverage = constant Vaverage = f (t)
Local .
Convective p ressure Gravzty F . . Now each of these mean forces has to be expressed as a
. . . . rzctzon
fi
znertza + znertza + orce + fiorce + fiorce = 0 function of the mean values and fluctuating values of the
fiorces fiorce velocity and the pressure. For this purpose, one considers:
-1 JT p(au
- + uau
- + v- au) dt
au + w-
7-2 Calculation of the Mean Forces
T o at ax ay az
= ~sT
T 0
(- o(p + pgz) + J.1'V2u) dt
ox
The mean forces are calculated as a function of the mean
values of velocities and pressure. For these calculations,
it is assumed that the order of mathematical operations
or, using the a and T notations and the rotational coefficients has no effect on the final result. In particular, integration
1J, (,and ~ (see Sections 6-1.2.2 and 6-1.3) during an interval of time T and derivatives with respect
to time or space can be interchanged.
- + -a (V
JT p (au 2
-1 - ) + 2(W1J - v() ) dt
r o at ax 2 7-2.1 The Constant Force
The gravity force depends only on the density of the
1
=+ T
JT (
0 -
a
ox pgz + ox
a (Jxx
a
+ oy Lxy
a
+ oz Lxz
)
dt elementary particle. The fluctuations of pressure are too
small to have a significant effect on the density. Hence, the
Similar equations are found for the 0 Y and OZ axes. gravity force is the same for laminar and turbulent motion.
Part 1 : Establishing the Basic Equations that Govern Flow Motion
74 The mean value of the gravity force is equal to this constant 7-2.2.3 The viscous force has a mean value which may
gravity force. Mathematically, this may be expressed be calculated by considering, for example, one of the second-
pg =T
1 r
Jo pg dt = pg T Jo dt = pg
1 rT
order terms, such as Jl.(0 2 ujox 2 ). Averaging this term
leads also to
~
p- = -
1 iT pOU
0 1
- dt = p - -
ot r
iT
o
u dt
7-2.3 The Quadratic Forces
Consider the component u = u + U 1
Squaring and
ot r o ot
7-2.2.2 Similarly averaging the pressure forces yields, Similarly, the mean value of uv is
for example:
op
--=--
av
ox ox and the mean values of UV and U V are zero.
1 1
Chapter 7: Turbulence: Mean and Fluctuating Components of Motion
T 0 ox ox ox OX
Considering each of these terms independently one obtains, This relationship, expressed as a function of the mean
since u and oujox are constant with respect to time, components of velocity and turbulent fluctuations, becomes
o a a
IT au
1
T u ox dt 0 =
au
u OX
ox (u + u') + oy (v + v') + oz (w + w') = 0
or
1IT ou'
T o u ox dt =
1 0 IT
u T ox o u' dt = 0 ou ov ow ou' ov' ow'
-+-+-+-+-+-=0
ox oy oz ox oy oz
-1 ITu,ou- dt -- -1 au- IT u,dt-- o
T 0 ox T ox 0 The averaging process, applied to oujox, gives
1 ITu' ou' 1 a IT u' 2 a u' 2 ~
T ox dt = T ox 2 dt = ox 2 = u' ox =f. 0 au au
0 0
ox ox
Introducing these values yields au'jox gives
and applied to
au (_au
pu-=p u-+ ~)
u- ou'
-=-=0
ou'
ox ox OX ox ox
Similarly, it is found that
Then the continuity relationship for the mean motion
becomes
ov ~)
8v ( u-+u-
pu-=p
OX ox ox
and so on.
Hence, the mean value of a convective inertia force with Con seq uen tl y,
respect to time is equal to the sum of the convective inertia
caused by the mean velocity and the mean convective ou' ov' ow'
-+-+-=0
inertia caused by the turbulent fluctuations. As far as the ox ay oz
mean value of the velocity alone is concerned, the convective
inertia terms have the same mathematical form as for the The mathematical form of the continuity relationship
case of a laminar motion. is the same for the mean motion as for the actual motion.
Part 1 : Establishing the Basic Equations that Govern Flow Motion
77
p(-au _au _au _ au ----:a;! , ou' ---:a,/)
+ u- + v- + w- + u - + v - + w - = -
a _
- (p + pgz) + Jl'V u
2_
(7-2)
at ax ay az ax ay az ax
Local Convective inertia Convective inertia Pressure and Viscous
inertia caused by the mean caused by the fluctuation gravity forces force
velocities velocities
oii
p( ot
- oii - oii
+ u ox + v oy + w oz
- oii)
= -
a - + pgz) + J.1V 2ii -
ox (p
(ou' 2 ou'v'
p ox + ay + Tou'w')
(7-3)
and shear stresses i. Hence, these new external forces to be their space derivatives is unaltered by a change in the
dealt with are: axes of reference. In particular:
Normal force: [ ] = (J- XX
(J XX -
--;z
pU u'v' = u'w' = v'w' = 0
_
= -p
au
+ 2f.l-- pu
--;z It is evident that isotropy introduces a great simplifi-
OX cation in the calculations. However, because of the
boundaries, the turbulence is not isotropic and the products
Shear stress: u'v', u'w' and v'w' may differ from each other. There exists
_ (au av) -puv
=-p+f.l-+-
---,, a correlation between u' and v', u' and w', and v' and w',
ay ax defined by the coefficients:
and so on. These new total external forces may also be u'v' uw v'w'
defined by a tensor of rank two similar to the first tensor (u'zv,z)lf2' (u'2w'2)1;2' (v'zw,z)l/2
defined in Section 5-5.2.
In practice the viscous forces caused by the mean These coefficients are equal to zero in the case of isotropic
velocity are very often negligible in turbulent flow in turbulence. Since the convective inertia forces caused by the
comparison with the other forces, and particularly in fluctuation terms are functions ofu' 2, v' 2, w' 2 , u'v', u'w', v'w',
comparison with the shear stresses caused by the fluctuation they may be expressed directly as functions of the coefficients
terms pu'v', pu'w', and pv'w'. of correlation which are dimensionless.
+~
oz [v(oii
oz +ow)_
ox u'w']
and two other equations which will be determined. Indicate
the advantage of this form of the Reynolds equation.
..[]
J1(o2 ujoy 2 ) increases near the boundary and becomes 81
particularly important in the case of a smooth boundary.
The mean velocity distributions in a pipe, given by
Fig. 8-3 and corresponding to different assumptions made
on the shearing stresses, illustrate these previous considera-
tions.
Figure 8-2 The shearing stresses caused by The quantitative study of the velocity distribution in a
turbulence decreases near the turbulent flow depends upon the assumption made on the
boundary, while the viscous force
distribution of the value of the shearing stress 1:. This is
increases.
the subject of the following sections.
Figure 8-3
Velocity distribution in a pipe.
--
--1-----+----x
-
-
~
--+----+---X
_.-/
82
TURBULENT FLOW;
MEAN MOTION IRROTATIONAL
EXCEPT IN THE BOUNDARY LAYER
8-2 Pressure Force. Secondary Effects pressure value. For example, it is seen in elementary
hydraulics that the hydraulic jump theory is developed
8-2.1 Paradox of Du Buat by equating the variation of momentum to the external
Considering the normal forces forces, i.e., the difference of pressure forces before and after
the jump (Fig. 8-5). To be more exact, it would be necessary
[a ] = - p ou
+ 2f1 ox - ""i2
pu to add to the pressure forces the difference in Jg (pu' 2 ) dh
(h is the water depth). This term is actually negligible.
It is seen that an additional force has been added to the However, this same factor is sufficient to explain why the
pressure force. This results in an increase of the average resistance of a body moving with velocity V in calm water
Chapter 8: Turbulence Effects: Modern Theories
,2 ,2
--+-+- pU pU
Figure 8-5
Turbulent fluctuation terms have to be added to the
pressure forces.
is different from the resistance of this same body when Finally, the Reynolds equations are reduced to :
stationary in a turbulent flow of the same mean velocity V.
This is the paradox of Du Buat. This phenomenon is caused op* o2 u ou'w'
by the difference of impulse of the turbulent convective 0= -Tx + f.l oz 2 - p ----a;-
inertia pu' 2 acting against the body in a manner similar to
op* ow' 2
pressure forces. o=- ---a; -P az
where p* = p + pgz. For a laminar flow between two
8-2.2 Turbulent Flow between Two parallel planes the Navier-Stokes equations would be
Parallel Planes written :
8-2.2.1 Consider the simple case of uniform steady
turbulent motion between two horizontal parallel planes,
as shown by Fig. 8-6. Since the mean motion is steady the
local inertia forces are zero : p(ouj ot) = 0; p(ovj ot) = 0;
p(owjot) = 0. Assuming that the mean velocity vector
is parallel to the two planes in the 0 X direction, the com-
ponents v and w along the OY and OZ axis respectively Figure 8-6 Turbulent flow between two
are zero and all the terms of the Reynolds equations where parallel planes.
those quantities appear are zero (see Section 7-5.2). Since
the mean motion is uniform, ouj ox = 0. It is then seen that
all the convective inertia terms are zero as in any case of
uniform flow.
Now consider the fluctuation terms. The derivatives of
u' 2 , v' 2 , w' 2 , u'v', u w
' ', v'w' with respect to x are zero since
the motion is uniform. The derivatives with respect to y are
also zero since the two planes are assumed to be infinite,
and the motion is two-dimensional. y
Part 1 : Establishing the Basic Equations that Govern Flow Motion
8-2.2.2 Integrating the second of the above equations only in name: they partly justify the use by engineers of the
with respect to z yields concept of hydraulic radius (Section 14-1.1). Indeed, the
definition of hydraulic radius is based on the assumption
p* + pw' 2 = constant that the shearing stress at the boundary is a constant. The
Let the pressure at the boundaries be p6. Since w' = 0 limitations of applicability of the hydraulic radius definition
must be known. A change in the secondary current pattern
at the boundaries, the mean pressure p* at any point of the
in a flow has an effect on the head loss which is not negligible.
flow is smaller than the mean pressure at the boundary
However, this effect is neglected in hydraulics because it is
p6 by the quantity pw' 2 not well known yet, and remains small.
8-2.2.3 If D0 is the mean velocity between the two planes,
it has been found experimentally that u' 2 !D6 < 0.01 and 8-3 Modern Theories on Turbulence
v' 2 !D6 and w' 2!D6 < 0.0025. Hence the difference: 8-3.1 The Unknowns in a Turbulent Flow
P6- p* w'z In Chapter 1, it was seen that problems in hydrodynamics
2-
"--'---~=-- consist of determining the four unknowns, u, v, w, and p.
(p/2)U6 - D6
For turbulent flow, the four unknowns are u, v, w, and p.
is always smaller than 0.0050 and is neglected. The pressure However, four other unknowns, u', v', w', and p' have been
distribution in a turbulent uniform flow is hydrostatic introduced, theoretically requiring four other equations
(at least within 0.5 %). (unless the fluctuation terms may be neglected). The
fluctuation values u', v', w', or some function of these values,
8-2.3 Secondary Currents
such as u'v', v'w', u'w', are expressed as functions ofthe mean
Such variations of pressure caused by the fluctuation values u, v, and w.
terms explain the origin of secondary currents in straight It is seen that p' does not appear in the Reynolds equation
channels and non-circular pipes. Secondary currents take which governs the mean motion because of the linearity
place when nonsymmetrical effects of the turbulent shearing of the pressure forces. Moreover, p' is usually very small
stresses exist in the flow, that is, each time that the boundary in comparison with p. p' should be taken into account
is noncircular. statistically only for some very special problems, such as
These secondary currents go from the zone of high in the investigation of the growth of wind waves under
shearing stress to the zone of lower shearing stresses as wind action. Although progress has been made in the
shown by Fig. 8-7. They have a tendency to equalize the statistical theory of turbulence, only isotropic and homo-
shearing stresses at the boundary. Hence they are secondary geneous turbulence is well described by use of random
functions. However, isotropic turbulence is an idealized
Figure 8-7 Secondary currents in a triangular pipe.
case never encountered as is the abstract concept of irrota-
tionality. Further investigation into nonisotropic turbu-
lence is necessary.
case of uniform flow parallel to a plane in the 0 X direction According to Prandtl, it is assumed that the fluctuation
(u = u(y), ii = 0, w = 0), is defined by the equality terms u' and v' are proportional to the difference in
pu'v' = -s(dujdy). Then the shearing stress [r] becomes velocity du which is equal to: du = (dujdy) dy, such that
[ r] = (f.l + s)(dujdy) instead of r = f.l(dujdy). [ r] is given u'v' = -l 2 (dujdy) 2 or Iu'l and Iv'l ~ l(dujdy).l is the "mixing
by a linear relationship. From this relationship it may be length." l may be physically considered as the length which
seen that the fluctuation term would act similar to the may be traversed by a lump of fluid perpendicular to the
viscous term. They are simply added linearly. However, mean velocity vector u. It is evident that according to this
they are of a different order of magnitude, that is s ~ f.1 definition, l is equal to zero at the boundary since a lump
and [r] ~ s(dujdy). of fluid cannot pass through the boundary.
Such a relationship gives a velocity distribution similar On the other hand, u'v' always has the opposite sign of
to that obtained in a laminar flow from the Navier-Stokes dujdy. If one considers a lump of fluid moving from the
equations. Since it would be necessary to consider that s boundary to the middle of the flow, v' > 0 and it is moving
varies with respect to space, the Boussinesq theory is a from a layer where u is smaller to a layer where u is larger.
failure. However, in some cases relative to the motion of It causes a slowing down of the motion, hence u' < 0.
atmospheric layer, s is approximately a constant and the Conversely, if one considers a lump of fluid moving toward
Boussinesq assumption is applied to obtain a result at a the boundary, v' is negative while u' is positive. Since
first order of approximation. u'v' is always negative, the shearing stresses caused by
turbulence, r = - pu'v', are positive, as is dujdy.
8-3.3 Prandtl Theory for Mixing Length If a velocity distribution has been considered such that
The mixing length theory has been introduced by dujdy is negative, it will be similarly found that - pu'v' is
Prandtl by analogy with the mean free path as it is defined always negative. Consequently, in any case, [r] has the
in kinetic theory of gases. It is the momentum transfer same sign as dujdy, which is emphasized by writing:
theory.
Consider the flow u = u(y), ii = 0, w = 0 parallel to [T] = plzl du ldu
the OX axis (Fig. 8-8). The mean velocities are u and dy dy
u + du at two points on a perpendicular to the boundary
defined by y = 0. In the general case,
_ du
[ T ] -f.l-+p du I-
12 1 - du
dy dy dy
Figure 8-8 Flow parallel to the OX axis.
-u
k In y + constant )
[ r ]) /2(1
Prandtl proposed:
p
1
ii = (
in a number of practical cases (wall, pipe, etc.). However, In the case of a smooth boundary, the term diijdy is no 87
they do not seem so successful when the flow is not uniform longer negligible in the boundary layer. Hence the head
(bend, divergent, etc.). It is to be born in mind that the loss is a complex intermediate function of ii, that is, a
phenomenon of turbulence is random, and therefore non- complex function of V.
deterministic. A solution to the problem of mathematical In the case of a laminar flow, [ r] is simply equal to
representation of such complexity is found in statistical r = Jl(dii/dy) and the head loss is a linearfunction of ii(ii = u),
mechanics, as introduced by Taylor, Von Karman, and hence a linear function of V.
Kampe de Feriet. This subject is beyond the scope of this
book. 8-4.2 Work Done by Turbulent Forces:
Dissipation Function
8-4 Some Considerations on the Loss It is evident that because of the turbulence the loss of
of Energy in a Uniform Flow energy in a turbulent flow is generally much greater than
in a laminar flow.
8-4.1 A Review of Elementary The mean value of the viscous forces per unit of volume
Hydraulics has been found to be
It has been seen in elementary hydraulics that the head
loss in a uniform flow is: J1'V 2 V = J1'V 2 V + 11'V 2 V'
where
1. Proportional to the mean value through a cross
section of the velocity V when the flow is laminar. J1'V 2 V' = J1'V 2 V' = 0
2. Proportional to its square value V 2 when the flow is The term J1'V 2 V is small by comparison to the kinematic
turbulent and the boundary is rough. forces caused by turbulent fluctuations. The mean value
3. A complex intermediate function of V (Vn) when the of the viscous forces caused by these turbulent fluctuations
flow is turbulent and the boundary is smooth is zero.
(1 < n < 2). If instead of considering the mean forces, one considers
The above result may be partly explained by the following the mean value of the work done by these forces, quite a
considerations. A part of the kinetic energy of the primary different result is obtained. Consider, for example, the
(or mean) motion of a turbulent flow is continuously mean shearing force Jl(au;ax). This force is equal to Jl(aujax)
absorbed to provide energy for the turbulent fluctuations. smce
The kinetic energy of these turbulent fluctuations is a
quadratic function of the fluctuating velocities. Since all
these fluctuations are finally absorbed by friction, the loss
Jle~) = JlG:) = o.
of energy in a turbulent flow is a quadratic function of the The work done by the force Jl(aujay) in a unit of time is
fluctuating velocities. (see Section 5-5.4)
On the other hand, the fluctuating velocities are roughly
linear functions of the mean velocities. It has been seen that JlG~Y dx dy dz
and by unit of volume: Jl(au;axf. The mean value of this
[r] = J1 dii _ pu'v' = J1 dii + p/ 2 dii Idii
1
88 The double product is zero: Explain the limitation of these assumptions. What is the
criterion for the pressure distribution on which such
2 au au' = 2 au !___ _!_ IT u' dt = 0 assumptions are based?
ax ax ax ax T 0
so that one finally obtains: 8.2 It has been found experimentally that
2+ J(au') 2
( au)
W=J1-ax l-
ax
where D is the distance between two parallel planes. Give
The second term is always positive, hence its mean value the expressions for [ r] and Prandtl's mixing length I as
is not zero. functions of y.
Moreover, u' is generally smaller than u, but the deriva-
tive of u' with respect to space (in this expression with 8.3 Using the Von Karman similarity rule
respect to x) is usually greater than the derivative of u with
respect to space (in this expression with respect to x). The 2 (dujdy) 4
first term (J1)(au;ax) 2 may often be neglected and W ~ r = pk (d2ujdy2)2
J1(au'jax) 2. Hence the loss of energy and the head loss are and the relationship (ap;ax) = constant along the centerline,
mainly due to the turbulent fluctuations. derive the following universal velocity distribution law
A similar result may be obtained by considering all the for a rectangular channel of width 2h
terms of the dissipation function <l> presented in Section
5-5.4 in which u, v, w are replaced by u, v, w, u', v', w'. Then
it is found that the mean value for <l> is the sum of two terms:
u_ = _u0 + -1 -
k
(r0)1/2{ [ (y)112] + (y)112}
p
In 1 - --
h h
<l>m and <l>t. <l> = <l>m + <l>t, where <l>m is a function of the
where u0 is the velocity at the centerline y = 0 and r 0 is the
mean values u, v, wonly, and <l>t, a function of u', v', w' only.
shear stress at the wall.
<l>m =
2
ax + ... + (au
J1[ 2 ( au) av) 2+ ...
ay + ~ J 8.4 It will be assumed that the velocity distribution in a
cylindrical pipe of radius R is given by the one-seventh
ii = _!_I
VO
Iff
vol
u d(vol) = _!_
A
If
A
u dA = V
Iff If
CONCENTRA TION CURVES
v = _!_I
VO vol
v d(vo!) = _!._
A A
v dA = 0
Iff If
Figure 9-1 Diffusion through a porous medium.
w= _!_I
VO vol
w d(vol) = _!_
A A
w dA = 0
6. This angle increases when the flow becomes turbulent;
it is then a function of the Reynolds number as defined where "vol" is the total volume of the porous medium. The
in Section 9-3.1. mean values with respect to space are written with two bars
instead of one to be differentiated from the mean value with
9-1.2 Definition of the Mean Motion respect to time {V) as used in studies of turbulent motions.
9-1.2.1 The simplest way of defining the mean velocity
consists of considering a unidimensional porous medium 9-1.2.2 For a more complicated pattern of the mean
as shown by Fig. 9-2. The mean velocity or "specific motion, where a variation of the mean value of the velocity
velocity" is the ratio of the discharge Q to the total area with respect to space also exists, the mean value of the
Figure 9-2
M ean uniform flow through porous medium.
y
Chapter 9: Flow in a Porous Medium: Law of Darcy
velocity vector has to be defined in an elementary volume elementary volume of a porous medium, it is found by a
dvol = dx dy dz of porous medium as follows: demonstration similar to that given in Chapter 3 that the
V= --
1 -Iff
dx d y dz ~x ~Y ~z
V dx dy dz
continuity relationship is:
au av
ax+ ay +
aw
az = 0
and along three coordinate axes:
That is, the continuity relationship for the mean motion
has the same mathematical form as for other kinds of
[Willv dxdydzffLx~y~z [u]v dx dy dz
~ 1
= flow.
W
92 The momentum equation of the mean motion in a porous with respect to space as was done for a turbulent flow
medium is obtained by averaging each of the forces with with respect to time.
respect to space, as has been done with respect to time in the Since the sum of the real value of the different forces
study of turbulent flow. This will be the subject of Section involved is always zero,
9-2. Both turbulent flow and flow through a porous medium
Inertia Gravity Pressure Friction
are strongly rotational as far as the real motion is concerned.
force + force + force + force = 0
However, their mean motions may be irrotational (see
Sections 8-1.2 and 9-2.2.2). An isotropic turbulent flow may The sum of the forces' mean values with respect to space is
by considered analogous to flow through an isotropic also equal to zero
medium. Turbulent flow through a porous medium is
Mean
studied by considering the mean values with respect to
value with [Inertia Gravity Pressure Friction]
both space and time such that
respect to force
+ force + force + force =0
IT ~X ~Y1 ~Z Iff
space
=
1
V=- v~~~~
T 0 &x&y&z
9-2.1.3 For a first approximation, the inertia forces are
neglected. The local inertia is neglected because the varia-
tion of the ground water table is usually very slow. From
9-2 Law of Darcy this point of view, unsteady motions through a porous
medium are usually studied as a succession of steady
9-2.1 Approximations motions (see Section 4-5.1). However, some special problems
9-2.1.1 First of all, it must be noted that for certain require the consideration of the local inertia, for example,
flows with a free surface through a very tine porous medium, perviousness of a rocktill breakwater to periodical gravity
the capillarity forces could have an appreciable effect on waves.
the flow pattern and the discharge. The convective inertia is also neglected. Since the velocity
For example, the rise of the free surface in an earth dam is usually very small, the square of the velocity and terms
of grain size near 0.1 mm is about 1 ft (30 em) (Fig. 9-3). which are functions of the square of the velocity (such as
the convective inertia forces) are of a second order of
9-2.1.2 Insofar as these capillarity effects may be magnitude in comparison with other terms (see Section
neglected, the momentum equation expressed as a function 4-5.2.1 ). The range of validity of such an assumption is
of the mean values is given by the same averaging operation studied further.
Figure 9-3
Capillarity effects in a flow through porous medium.
Chapter 9: Flow in a Porous Medium: Law of Darcy
~ ~1 ~
X Y z
Iff ( dxdydz
a(p- +
- - apgz)
--
X
+ J.i'V 2 u) dx dy dz = 0
_ a _
v = K ay (p + pgz) = Kay
ar*
~ X
1~ Iff
Y Z dxdydz
[-grad (p + pgz) + J.1V' V] dx dy dz
2
or vectorially:
~ X
1~ Iff . -
Y Z dxdydz
grad (p + pgz) dx dy dz
Stokes equations are no longer valid, since from a micro-
scopic point of view, a flow passing through the very fine
channels of a porous medium, like porous china, would
=-grad (p + pgz) probably require a study based on molecular agitation. This
subject is relevant to the kinetic theory of liquids.
Now consider the viscous forces. Since they are linear, In a nonisotropic porous medium, K has different values
it is reasonable _!o expect that they are proportional to the -Kx, Ky, Kz-along the three components axes OX, OY,
mean velocity V as long as there are no quadratic effects and OZ, respectively.
caused by the convective inertia and tur~ulence. Hence
it is written that they are proportional to J.lV such that
9-2.2.2 It is important to note that such a mean motion
~X ~Y ~z
1 Iff dx dy dz
J.lV' 2 V dx dy dz
k
V
= -J.l=
defined by the law of Darcy is always irrotational.
Introducing the value u, v, wgiven above, it is easy to
verify that au;ay - av;ox = 0, since
where k is the permeability of the porous medium, k is an
empirical function of the void coefficient and grain size.
K = k/ii is the hydraulic conductivity which measures the
!___ [~
ay 1-i ox
op*] _!___ax [~1-i op*]
ay
=0
Part 1 : Establishing the Basic Equations that Govern Flow Motion
94 z
AT TH E FRE E SURFACE
Similar demonstrations may be done for the two other 9-3 Range of Validity of the Law of
conditions given in. Section 2-6.2. However, if the flow Darcy
through the porous medium is turbulent, it is necessarily
rotational and hence cannot have its mean motion defined 9-3.1 Effect of Reynolds Number
by a potential function. 9-3.1.1 The permeability k is a function of the porous
medium characteristics and the Reynolds number. The
9-2.2.3 The velocity potential function for a laminar Reynolds n~mber of a f!ow through a porous medium is
flow through a porous medium is usually defined_ by defined by V6/v where V is the mean or specific velocity
u = - K(o</>/ox), iJ = - K(o j 8y), w= - K(o</>/oz), or V = as previously defined, and (j is the diameter of a grain.
- K grad . Substituting these values into the Darcy The diameter of a grain is easily known while the" diameter
equations gives of the channels," as used for pipe, would be difficult to
define. This process assumes that there is a simple linear
::1 ..1,
u'f' 0" relationship between these "channel diameters" and the
= - K - (p + pgz)
~ ~
u=- K -
ax ax grain size. However, in a porous medium made of a large
grain size distribution, the small particles have a tendency
and two similar equations for v and w. From these, it to reduce the size of the "channel." The "channels" have
~easy to see that is equal to the piezometric head: the same order of magnitude as the smallest particles.
p* = (p + pgz) = . Sometimes is also defined by: Hence, it is more exact in this case to define the Reynolds
[(pjpg) + z] = p*jpg and u = - Kpg(a;ax), etc. Figure 9-4 number with the help of the smaller grain sizes. The
illustrates the value and the physical meaning of corre- "characteristic diameter" (jc may be considered empirically
sponding to such a definition. The velocity potential as the average size corresponding to the lowest 10 percent
function is a constant along the sides of the dike and limit.
decreases linearly with z at the free surface inside the dike.
Hence a constant Llz corresponds to a constant value for 11. 9-3.1.2 Although the velocity in a porous medium is
This property will be used in Section 11-6.3 for devising very small, the derivative of velocity with respect to space
a graphical method of investigation. is large. It is easy to recognize this since the actual paths
Chapter 9: Flow in a Porous Medium: Law of Darcy
Then the coefficient b for the same void coefficient and the same value of the Reynolds number (Fig. 9-5) (see Section
same grain size distribution is a function of the roughness 13-5.1). This empirical law has been established experi-
of the rocks. A similar phenomenon has been found in mentally for a range of Reynolds numbers between 10 2 and
elementary hydraulics in the case of turbulent flow in a 10 5 and for small variations of the void coefficient 8, such as
rough pipe (see Section 13-4.3). encountered in sand or quarry run used in rubble mound
'
"" ~
Figure 9-5
Drag of a rough sphere vs 10
""' "'
--
Reynolds number SMOOTH SPHERE-,
~ ROUGH SPHERE l
.............. l
..
'
,___ j
-2 -I 2 3 4 5 6
10 10 10 10 10 10 10 10
V8c
-v-
Part 1 : Establishing the Basic Equations that Govern Flow Motion
breakwaters and rockfill cofferdams. For a laminar motion For flow through a porous medium the mean motion
with a negligible convective inertia (Re < 1), ~ith respect to space is dealt with. The two unknowns are
V(u,v,w) and p.
24
ex= vbjv" In all cases (ideal fluid, laminar flow, turbulent flow,
flow through porous medium) the two unknowns-velocity
Hence the hydraulic conductivity is found to be equal to and pressure, real or mean with respect to time or with
respect to space-are obtained by the continuity relation-
K = ~ pgb;a3 ship and the momentum equation.
pl2p(l - a) 2
Convective
Pressure Viscous inertia caused
Inertia gravity friction by turbulence
du av*
Perfect fluid: Eulerian equation --
p dt ax
Laminar flow: Navier-Stokes du av*
equations p dt ax
+ f-l'Vzu
and as a function of the mean velocity with respect to space average flow L = 100ft. One wants to build a scale model 97
for a flow through a porous medium: ofthis porous medium at a scale A = 1/10 such that a = 1 ft 2
an av aw (929 cm 2 ) and I = 10 ft (305 em), and with the same void
-+-+- =
0 coefficient e. One wants the discharge to obey the rule of
ax ay az similitude of Froude, i.e., q = A512 Q under a head Llhmodei =
9-4.3 The Momentum Equations Llhprototype x ),. For this purpose the grain size of the model
The momentum equations are written (in the table 6"' will be related to the grain size of the prototype c5 P by the
opposite) along the OX axis only for the four cases to relationship 6"' = KA6P. Determine the value of K in the
be considered. Since similar terms are found in these case where H = 100 ft (3047 em) c5 = 0.024 in (1 mm),
four equations, similar methods of integration may be and e = 0.40.
used. Some of them are only valid after some approxi-
mations or some assumptions have been used to simplify REFERENCES FOR PART ONE
the basic equations. For example, neglecting the turbulent
fluctuation terms and the viscous term, a turbulent flow Daugherty, R. L., and Ingersoll, A. C., Fluid Mechanics, 5th ed.
McGraw-Hill, New York, 1947.
behaves as a perfect fluid. Hence, in order to simplify the
writing and for more generality, only the notation V(u,v,w) Goldstein, S., Modern Development in Fluid Mechanics, Vol. I.
Oxford University Press, 1938.
and p are used in the following chapters since it is under-
Hunter, R., Editor, Advanced Mechanics of Fluids. Wiley, New York,
~ood that V and p means V and p for turbulent flow and
1959.
V and p, respectively, for flow through a porous medium.
Lamb, H., Hydrodynamics, 6th ed. Dover Publications, New York,
1945.
PROBLEMS Landau, L. D., and Lifshitz, L. D., Fluid Mechanics. Pergamon
Press, London, 1959.
9.1 Calculate the total flow discharge through a porous Li, W. H., and Lam, S. H., Principles of Fluid Mechanics. Addison-
medium of total cross section A = 1 ftl (929 cm 2 ) and length Wesley, Reading, Mass., 1964.
in the direction of the flow l = 3 ft (91.4 em) as a function Milne-Thomson, L. M., Hydrodynamics. Macmillan, New York,
of the head. The significant grain size diameter is c5 = 0.07 in 1960.
(0.3 mm) and the void coefficient is e = 0.40. One will also Owczarek, J. A., Introduction to Fluid Mechanics. International
make use of Fig. 9-5 for determining the head loss coefficient. Textbook, Scranton, Pennsylvania, 1968.
Determine the head under which the law of Darcy no longer Planck, M., The Mechanics of Deformable Bodies. Macmillan,
applies, and the head under which the turbulence appears. London, 1932.
Repeat the same calculation when the porous medium is Prandtl, L., and Tietjens, 0. G., Fundamentals of Hydro and Aero-
composed of two successive layers: I= 1.5 ft (45.7 em) mechanics. McGraw-Hill, New York, 1934.
and c5 = 0.12 in (0.5 mm); I = 1.5 ft (45.7 em) and c5 = 0.024 Rouse, H., Elementary Mechanics of Fluids. Wiley, New York, 1946.
in (0.1 mm); and three successive layers of length I = 1 ft Sabersky, R. H., and Acosta, 1. A., Fluid Flow-A First Course in
(30.5 em) each and c5 = 0.024 in (0.1 mm), 0.072 in (0.3 mm), Fluid Mechanics. Macmillan, New York, 1964.
and 0.12 in (0.5 mm) respectively, of same void coefficient Serrin, J., "Mathematical Principles of Classical Fluid Mechanics."
(kinematic coefficient of viscosity v = 1.076 x 10- 5 ftl /sec In Encyclopedia of Physics (S. Fligge, Ed.), Vol. 8-1. Springer,
Berlin, 1959.
(0.01 cm 2 /sec).
Streeter, V., Editor, Handbook of Fluid Mechanics. McGraw-Hill,
9.2 Consider a flow through a porous medium with a cross New York, 1961.
section A = 100 ft 2 and a length in the direction of the Streeter, V., Fluid Dynamics. McGraw-Hill, New York, 1948.
PART TWO
Son1e Mathen1atical
Treatn1ents of
the Basic Equations
Chapter ro
101
= -
op*
(- dx + -op* dy + -op* dz )
ox oy oz
!02 the left side of the above equation can be written in the form
p(u du + v dv + w dw) = pd
2 yz)
(vz + v2 + wz) = d(P2- 10-1.2 The Case of Irrotational Flow
In the case of an irrotational flow, one has seen (Section
The right-hand side is the total differential of p*. Hence, 2-6.3) that the velocity vector could be defined by a potential
the momentum equation becomes function rjJ such that V = grad , and
or/J a a v = a
d(p ~2 ) = -dp* u = --
ox v=-
oy w=-
oz as
Specifying that the above equation applies along a stream- Let us consider again the momentum equation in the X
line only and that p* = p + pgz, the final result is direction, and replace the velocity components u, v, w, by
their expressions in terms of rjJ (see equation 6-1):
;~(PY; + p + pgz) = o o r/J
2 (o o
2 o 82
o 82 ) op*
p oxot + p ox ox + oy oxoy + oz oxa-;
2 = ox
When integrated this becomes
vz Since
pT + p + pgz =constant
a ( a v2
ax pat- + p2- + p*
)
- = 0
p -V + p + pgz + p Jav
at dS pat+ p2 + p* F(y,z,t)
2
=
2 = constant
Differentiating with respect to S gives Similar equations are obtained in the Y and Z directions so
that the function F is a function of timet only. It is important
-a (pV av= 0 to point out that this equation is valid everywhere in the
2
- + p + pgz ) + p -
as 2 at fluid and in any direction, particularly along a streamline.
Chapter 10: The Bernoulli Equation
Indeed, taking the derivative with respect to S yields applies both to steady slow motion and to the case of no 103
motion. However, in the latter case the law of hydrostatics
fP<P
P at as
a (P 2V
+ as
2
+ p*
)
= 0 is physically exact, in additio"n to being mathematically
exact, since the case of no motion implies no friction.
But, since V = a;as, then this can be written as
10-2.1.2 A similar law is obtained for a uniform steady
flow of a real fluid, i.e., with friction forces. Consider the
uniform flow studied in Section 6-2.1.1. The pressure
A number of simple cases are reviewed (vide infra), with a distribution obtained was: p = Pa + pgz cos a, where Pais
physical interpretation. the atmospheric pressure.
The isobars or lines of equal pressure are inclined at an
angle a with respect to a horizontal plane. Therefore,
10-2 Simple Cases and Physical buoyancy exerted on any body in such a flow, such as on a
Interpretations rock deposit on the bottom, is exerted at an angle a with
the vertical.
10-2.1 Slow-Steady and Uniform-Steady For a number of practical cases of flow with a free surface,
Motions a is very small and cos a may be considered equal to unity.
Therefore, the pressure distribution in a free surface uniform
10-2.1.1 As the motion is steady, the local inertia terms
flow is most often hydrostatic.
are zero; because the motion is slow, the convective inertia
This hydrostatics law is considered to be accurate
terms may be neglected; (and because the fluid is perfect,
enough even for a nonuniform and nonslow motion when
the friction forces are zero). Hence, the momentum equation
the curvature of the paths is small. The calculation of a
is reduced to an equality of applied forces-pressure and
backwater curve as seen in elementary hydraulics is usually
gravity, mathematically expressed as
based on such an assumption, even though this approxima-
ap = o tion is not always specified.
ax Figures 10-1 and 10-2 illustrate two cases in which the
convective inertia has a nonnegligible influence on the
av = o
ay
Figure 10-1 Pressure distribution is greater than
o(p + pgz) = 0 that given by hydrostatic law.
oz
---=------'---
HYDROSTATIC PRESSURE
As the axes 0 X and 0 Y are horizontal, the first two
equations show that the pressure is a constant in a hori-
zontal plane. The third equation gives
p + pgz = p* =constant
It is seen that p varies linearly with the distance from the
free surface. The law of hydrostatics is recognized; that is, BUCKET
hydrostatics could be considered as a limiting case of
steady slow motion of an ideal fluid. Therefore, the same law
Part 2: Some Mathematical Treatments of the Basic Equations
104
REAL PRESSURE
(see Section 6-1.2.4). Dividing by w = pg yields the total
DISTRIBUTION
head H, which has the dimension of a length
HYDROSTATIC
PRESSURE
vz p
-+-+z=H
2g w
The total head is the sum of the velocity head V 2 j2g,
the pressure head pjw, and the elevation head z. The value
given by a piezometer, (pjw) + z, called the piezometric
head (see Fig. 10-3).
It is seen that the variation with respect to space of the
Figure 10-2 Pressure distribution is smaller than
that given by hydrostatic law. total head H in an irrotational motion is zero: grad H = 0.
Such an equation expresses the conservation of energy of
an elementary particle of fluid as a sum of its kinetic energy,
pressure distribution, and conversely, the pressure distribu- pressure energy, and potential energy. It is emphasized that
tion has an influence on the flow pattern. The effect of flow the Bernoulli equation is valid in any direction for an irrota-
curvature is studied in Section 10-2.4. tional motion, i.e., along a path as well as along a normal to
a path. It is noticed, also, that the velocity V and pressure p
refer to the local value of the velocity and do not refer to
10-2.2 Slow-Unsteady and Uniform- the mean velocity and mean pressure in a cross section of a
Unsteady Motions of a Perfect Fluid stream tube. The cases in which Bernoulli's equation can
The rotational convective inertia terms are nonlinear. be, approximately applied to stream tube are discussed in
Therefore, a slow motion may mathematically be considered the following section.
irrotational and V(u,v,w) may be defined by a potential In engineering practice, the pressure along a boundary
function . The quadratic term p V 2 /2 may be neglected. may be determined by this equation when the velocity
Therefore the momentum equation is field is given, and the flow is approximately irrotational,
as encountered in short, convergent structures (see Section
8 2-5). In a steady flow, if 8Hj8S is positive, it is because of
pat + p + pgz = f(t)
the action along the path dS of an external force such as a
pump. If 8Hj8S is negative, it is either because of the action
which is valid in any direction throughout the fluid. The along the path dS of an external force, such as a turbine, or
derivative of the function f(t) with respect to space is zero. because of friction force.
The variation in H along S measures the action of
10-2.3 Steady Irrotational Motion of a turbo machines or the head losses in the dimension of a
Perfect Fluid length.
10-2.3.1 The momentum equations in the OX, OY, OZ
directions of steady irrotational motion of a perfect fluid
give 10-2.3.2 In a rotational flow H varies from one streamline
to another streamline, whereas His the same for any stream-
v2 line of an irrotational flow. Figure 10-4 as compared to
p2 + p + pgz = constant
Fig. 10-3 illustrates this point.
Chapter 10: The Bernoulli Equation
105
TOTAL HEAD
FREE
SURFACE
2
v
29
VELOCITY
HEAD
Figure 10-3
In an irrotationaljiow, the total head is
a constant at any point.
DATUM
10-2.4 Pressure Distribution if! a 10-2.4.2 Now the general case ofnonnegligible curvature
Direction PerpendiclJlar to the is studied. Consider an infinitely small, curved, two-
Streamlines dimensional stream tube as shown by Fig. 10-5, and an
elementary mass of fluid p dR dS in this stream tube.
10-2.4.1 In an irrotational flow, the derivative of p is
Because the motion is in the direction of the stream tube,
known in any direction by applying the Bernoulli equation
this elementary mass is in equilibrium in a direction normal
to the velocity field.
to the streamline and is under the action of its inertia and
In a rotational flow, the Bernoulli equation gives the
applied forces.
derivative of p along a streamline as a function of the deriva-
Its inertia gives rise to a centrifugal force equal to
tive of V, but does not give any indication of the derivative of
p dR dS V 21R, where R is the radius of curvature of the
p in a direction perpendicular to the streamlines. However,
streamlines. The applied forces are the difference of pressure
both of these are of equal importance in engineering
forces acting on the two streamlines
practice.
It has been seen that the pressure distribution in a
uniform flow is hydrostatic. This hydrostatic law is again + op dR) dS - p dS = op dR dS
(p
valid when the path curvature is small. oR oR
Part 2: Some Mathematical Treatments of the Basic Equations
I06
VELOCITY
HEADS
PIEZOMETRIC HEAD
HEADS
Figure 10-4
PRESSURE
In a rotational flow, the total head changes
from one streamline to another.
and the gravity Integrating this equation along dR permits the calcula-
tion of the pressure distribution from the velocity fields
pg dR dS cos IX
along a curved boundary, as on a bucket of a spillway, for
From the diagram it is observed that example.
If R tends to infinity, o(p*fw)foR tends to zero and
(ozfoR) dR oz p* = constant, as has been found in the case of uniform
cos IX = dR = oR flow.
When the centrifugal force is equated to the applied forces Figure 10-5 Notation.
and the resulting equation is divided by the volume dR dS + ~dR
it gives p ~ R
V2 o(p + pgz)
p-=
R oR
or
RADIUS OF
CURVATURE
that is
Chapter 10: The Bernoulli Equation
*) *)
(real fluid). OZ at an
p + pgz cos a = constant
angle a with the
vertical
(P 2q + Pz Vz dAz- ( P 2 Vi + P1 V1 dA 1
(P ~2 + p*)v dA
is an energy flux or power.lt is seen in elementary hydraulics
that the energy flux through a cross section A is given by
the expression
[v + pP + J
Figure 10-6 Mean velocity in a cross section.
2
1
H.P. = 550 z VA
29
in the ft-lb-sec system, where H.P. is in horse power.] Finally, V' is caused by turbulent fluctuations and is a
measure of the deviation of V from its time average. It
10-3.1.2 Across a section such as BB', the velocity V can varies both in time and direction, and is such that
be written as the vector sum (see Fig. 10-6)
V = U + U' + V' V' = s: V' dt = 0
U' = ~ LT(V - U) dt 10-3.1.3 Across the section the pressure p* = p + pgz will
also vary with time and space and one can write similarly
and
If
A
U'dA = 0
p = p + p' + p"
However, the variations p' and p" are neglected.
Chapter 10: The Bernoulli Equation
with a = 3a 2 + 3x 2 For laminar flow in a circular pipe, 10-4 Limit of Application of the Two
there is no turbulence, V' = 0, and 3x 2 = 0. In the case of Forms of the Bernoulli Equation
a paraboloid velocity distributions, as found in elementary 10-4.1 The Two Forms of the Bernoulli
hydraulics, it is found that 3a 2 = 1. For turbulent flow Equation
3x 2 ;;:; 0.05 and 3a 2 ;;:; 0.05 to 0.01 (see Section 13-4). In the case of steady flow without friction, two forms of
Finally, consider the local inertia term the Bernoulli equation are almost identical: the first one is
_!_ IT JJ Js P avat v dA dt ds v2 p
+ - + z = constant
T Jo A
-2
g pg
Simplifying as previously seen, it becomes throughout the fluid, and the other one, valid along a
f
P s
A a(u2 /2)(1 + -;;z + x2) dS
at
=
P
A(1 + ~)u au
3 at
stream tube, is
(1
u2 p
+a)-+-+ z =constant
2g pg
Strictly speaking, neither of these equations is valid
10-3.2 Practical Form of the Bernoulli
in any real case because the conditions required for estab-
Equation for a Stream Tube
lishment of the Bernoulli equation can only be approxi-
Taking account of the above correction factors, and mated. However, they are essential in many cases. In
dividing by Q = U A, the generalized form of the Bernoulli such cases, it is important to remember the following
equation for a stream tube is assumptions:
The first form of the Bernoulli equation is valid for irrota-
(p ~~ (1 + 11.2) + P2 + pgz2) tional flow, that is, in convergent flow through short
structures. V is the local velocity. In case of turbulence, V
g1 Js 8t
is not much influenced by the vorticity in thin boundary
( u~ p2 ) (ufg + P1 ) au dS
2g + pg + z 2 - 2 pg + z 1 = layers.
The second form of the Bernoulli equation is valid for
In the case of a uniform flow in a pipe of length L, the local unidimensional flow where the motion is rotational, but the
Part 2: Some Mathematical Treatments of the Basic Equations
such as circular or square. But in the case of turbulent flow, measurement is locally influenced by a complex flow I I I
the value of head loss cannot be obtained by theory and pattern.) However, in the case of nonuniform flow, it is
must be measured by experiment. more difficult because Ui/2g is different from U~/2g; in
In this case, the Bernoulli equation permits us to define particular, oc 1 is different from oc 2 and they are unknowns.
the value of the head loss !):.H between two considered Even if they are considered as known, the definition of such
cross sections by the difference between the total heads a head loss involves lack of accuracy as evidenced by Fig.
at these points 10-9. Moreover, it is impossible to separate the value of
[ (1 + oc 1 ) -uz
g
p*] [
1 + ___!_
pg
- (1
g
p*]
+ oc 2 ) -uz2 + _2
pg
= !):.H
head losses caused by a succession of close sudden changes.
A linear addition of the various head losses cannot be made
because of the interaction between the effects of successive
This does not present any difficulty as long as the flow is sudden changes.
uniform. In this case
10-5.3 Head Loss in a Free-Surface
!):.H = pifpg- pjjpg Flow
It is interesting to note that any change in a free surface
10-5.2 Head Loss at a Sudden Change flow gives no extra head loss. Indeed, the initial upstream
in Flow level and downstream level are always the same provided
A sudden change in a uniform pipe, such as caused by a the considered cross sections are far enough from the
diaphragm or a bend, has an effect on the flow velocity discontinuity as is illustrated by Fig. 10-10. The increase of
distribution and the head loss at a great distance down- head loss in one place is always compensated by a decrease
stream. The head loss due to this change may be obtained of head loss in another place. Hence, the head loss definition
by extrapolating the pressure lines as shown by Fig. 10-8. of a discontinuity or sudden change in a free-surface flow,
(A pressure given by a piezometer located near a dis- such as that caused by a grid, must be specified by the
continuity, a bend or an intake has no value in evaluating relative location of two cross-sectional planes between
the head loss because the flow is not uniform. The pressure which the head loss is considered.
0>
a.
......
0..
0
I
-- -- --
--
<{
Figure 10-8 w t-------t----+---.0. H = constant
-- --
I
The head loss 11H caused by a singularity or
w
sudden change in a uniform pipe 0::
::)
is well defined by extrapolation. (f)
(f)
w
0::
0..
EXPERIMENTAL POINTS
DISTANCE S
Part 2: Some Mathematical Treatments of the Basic Equations
112
o>
o._
'
Q_
Cl ll.H =?
<(
Figure 10-9
',--+t ----
w
::r:
w The head loss caused by a sudden change
Q: in a nonuniform flow is subject to
::::l
various interpretations.
EXPE~
(J) I
(J) zw l
w wl!>l ' ........
0:: OZI
a.. ~1
enol
DIS TAN CE S
However, in case of sediment transport, the statement The head losses for the same value of the mean velocity
may not hold true as illustrated in Figs. 10-11 and 10-12, as are different for steady flow, accelerated flow, and de-
the sediment tends to accumulate upstream, proportionally celerated flow. This may also be noticeable for a discon-
to the head loss. tinuity such as the bottom orifice of a surge tank, where the
flow pattern is influenced by an instability phenomenon.
10-5.4 Effect of Local Inertia upon However, owing to lack of experimental data, unsteady
Head Loss flows are often studied with a head loss given by an empirical
By having an effect on the velocity distribution in a pipe, law experimentally obtained in the case of steady flow. The
and on the corresponding shear stress, local inertia has an studies of tide in estuary and storm surge are based on
important effect on the value of the head loss (see Fig. 10-13). such assumptions. As in the case of steady flow, the
Head loss in a given flow at a given time cannot be Bernoulli equation may be used to define the head loss in an
considered theoretically as equal to the value of the head unsteady flow, but even more difficulties are encountered
loss of the steady flow, which would have the same instan- in determining the head loss experimentally in unsteady
taneous value of mean velocity. flow than in steady flow.
Figure 10-10
A discontinuity in a free surface flow does not
change the total value of the head loss L1z 12 .
Chapter 10 : The Bernoulli Equation
II3
Figure 10-11
Small dam .
10-5.5 An Application to Surge Tank v;;2g is negligible as V, in the reservoir is very small.
VV 2g is also very small and is usually neglected. Moreov~r,
Because of its importance in engineering practice, an
Pa = Pb = atmospheric pressure. The head loss term m-
example of the generalized Bernoulli equation for un-
cludes the head loss at the entrance of the gallery, the head
steady motion is given. The case of a surge tank in the case
loss in the gallery ilH = LfV 2 /2gD, and the head loss due
where the discharge in the penstoke is zero is analyzed
to the bottom diaphragm of the surge tank KVJ/2g. The
(Fig. 10-14). The application of the Bernoulli equation
head loss in the surge tank is usually negligible as is VV 2g.
between points a and b gives
(Lis the length of the gallery andfthe head loss coefficient.)
The local inertia term is usually small enough in the
reservoir and in the surge tank to be neglected. It is taken
into account in the gallery only. Hence,
Figure 10-12
Large dam .
Part 2: Some Mathematical Treatments of the Basic Equations
JL ) V2 L dQ
Z =(D + ka-b 2g + gf dt
In view of the results, explain the shape that a drop of rain
where ka _ b is a friction coefficient for singularities between and an air bubble in water take.
a and b. The continuity equation is Q dt = F dZ, F being
the cross section at the free surface of the surge tank. 10.4 Demonstrate that the Bernoulli equation is valid along
a streamline using the two following fully developed forms
of the convective terms of the momentum equation
PROBLEMS
ou +v -au +w-
( ox
pu -
au)
10.1 The velocity around the limit of a circular cylinder is oy oz
given by the equation V = 2U sin ewhere U is the velocity and
J
at infinity, and where the pressure is Poo. Determine the
pressure distribution around the cylinder. {:x (~2 ) + 2(ryw - (v)
Figure 10-14 Surge tank. 10.5 Two adjacent tanks have horizontal cross sections S 1
and S 2 , respectively. The difference of level between these
two tanks at time t = 0 is h 1 - h2 = h. An orifice of cross
section A is open between the two tanks at time t = 0. The
discharge coefficient of the orifice is 0.6. Give the expression
for the time T after which the level in the two tanks is the
same.
A B c D SURGE TANK I 15
t
60 II T T
5011
l
l
4011
1" j_ lOft
I
.. ' H
I 1 T I
M N Figure 10-16
Figure 10-15
2. Give the period of oscillation of the motion in the
is maintained at gallery. (P will be neglected for this calculation.)
A: zA =60ft (18.28 m) B: z 8 =40ft (12.19 m) 3. Give the amplitude of oscillation of z in the surge tank
C: zc = 50ft (15.24 m) D: zc = 10ft (3.04 m) in the case for which the initial discharge Q, = f~ is
suddenly stopped to a zero value and to a smaller value
respectively. The pipe between A and B is 10 in. (25.40 em) in Q; = fW~. (P will again be neglected.) Explain
diameter and 3000 ft (914.4 m) long. The pipe between C qualitatively the influence of P.
and D is 12 in (30.48 em) in diameter and 6000 ft (1828.8 m)
long, and finally, the connecting pipe MN is 5500 ft 10.8 Establish the basic equations of motion for unsteady
(1676.4 m) long, M being 1000 ft (304.8 m) from reservoir flow in parallel pipes and in series.
A, and N being 2000 ft (609.6 m) from reservoir D. The
friction coefficients f of these pipes are 0.20 for the lO- 10.9 Establish the equation of motion caused by the sudden
in. and the 12-in.-diameter pipes, and 0.224 for the pipe opening of a gate for a manifold such as that shown in
between M and N. The diameter of the pipe MN is such that Fig. 10-17.
the discharge through MN is 1.2 ft 3 /sec (33,980 cm 3 fsec).
Determine the discharges between AM, MB, CN, and ND
and the diameter of the pipe MN. Figure 10-17
and that theoretically, both may be found directly from the practice the stream function is mainly used in two-dimen- I 17
momentum equation and continuity relationship. However, sional flow; only this case is analyzed in this book.
in many cases the methods under study provide a knowledge The stream function may be defined by any one of its
of the relative velocity distribution from the velocity field characteristics and then the other characteristics may be
calculated from the continuity principle, and an assumption deduced from this chosen definition. As was done for the
such as that of irrotationality only. velocity potential function, the stream function will be
To calculate the absolute value of the velocity requires defined first by the velocity components.
a second step. This second step is simple when the absolute
value of the velocity at one point or at one boundary is 11-2.1.2 The stream function is a natural outcome of
known. the continuity relationship
Then, in a third step, the pressure distribution is deter- au ov - 0
mined by application of some of the relationships between ax+ oy-
V and p, which have been established in Chapter 10, such
as the Bernoulli equation. Indeed, consider a function t/J(x,y,t) = constant such that
This chapter deals with the problem of the determination ot/J ot/J
of the velocity field by some analytical methods of particular U=-
oy v = - ox'
importance. These analytical methods are based on the
use of two mathematical tools that allow a concise descrip- From the continuity relationship given above, it follows
tion of the complete flow pattern. They are the stream that
function and the velocity potential function. a ot/J a ot/J _ 0
ax ay- ay ax=
which shows that t/1 always satisfies the principle of con-
11-2 Stream Function tinuity; in other words, the existence of t/1 implies that the
continuity relationship is satisfied and conversely the
11-2.1 Definition continuity equation implies the existence of a stream
function t/1.
11-2.1.1 The stream function is a mathematical device to
describe a flow pattern concisely by its streamlines. The 11-2.2 Stream Function, Streamlines,
stream function may be used to calculate any kind of and Discharge
flow of incompressible fluid: rotational or irrotational;
11-2.2.1 Now it is shown that such a function t/J =
steady or unsteady; two-dimensional or three-dimensional;
constant is not only the equation of one streamline but of
laminar or turbulent; slow or nonslow motion. However,
any streamline of the considered flow. This is performed
in the case of turbulent motion, the stream function is
by a simple change of the constant value for t/1.
intended to define only the mean motion with respect to
For this purpose, consider the streamline equation
time, i.e., the mean velocity vector V. It may also be used to
dxfu = dyjv (see Section 1-2.2), which may be written
define the mean motion with respect to space of a flow
u dy - v dx = 0. Introducing the value of u and v as
through porous medium whatever the value of the Reynolds
functions of t/1 yields the equation of streamlines in terms
number, i.e., for turbulent flow as well as for laminar flow.
of stream function
Although the stream function may theoretically be
defined and used for three-dimensional motion, its calcula- ot/J ot/J
-dx + -dy = 0
tion is complex and its use has been limited. Hence, in ax oy
Part 2: Some Mathematical Treatments of the Basic Equations
that is, 2( = V2 t/J. Hence, an irrotational motion for which function permits the transformation of a system of two 119
( = 0 is defined by a stream function t/J, which is a solution equations with two unknowns u and v into one equation
of the Laplace equation V2 t/J = 0. In other words, V2 t/J = 0 of higher order with only one unknown.
defines an irrotational motion which satisfies the continuity When boundary conditions are introduced, this equation
principle. It may be easily verified that the example of gives the theoretical value of t/1 after successive integrations
uniform flow defined by a stream function given in Section from which u and v are afterward obtained by simple
11-2.2.2 is irrotational. differentiation.
122 Table 11-1 Stream and potential functions Therefore, a velocity potential function exists.
V2 t/J = 0 V2 c/J = 0
Also
X= r COS 8
streamlines are represented by a family of rectangular y = r sine
hyperbolas that represent a flow toward a plate per-
pendicular to the incident motion, as shown in Fig. 11-4. ox ! ox= -sine
Such a motion is irrotational because u = x, v = - y, and or= cos e roe
1 oy
~~=sine --=cos
roe
e
Chapter 11 : Flow Pattern, Stream Function, Potential Function
123
Figure 11-4
Flow toward a plate
_ 1 [ot/J
-- -- - ox]
oy+ot/J _-
- - 1 ot/J
- Vo = -;:
1 [aq;OX (- r . 8) + aq;oy (r
Sill COS 8)
J
r oy o8 OX o8 r o8
and = ~ [aq; ax + aq; oy] = ~ aq;
r ox o8 oy o8 r o8.
V0 = - -
. 8 - -ot/J COS 8
ot/J Sill
oy ax Finally
ot/J 1 aq;
ot/J ay ot/J ax ot/J Vo =--=--
or r o8
oy or OX or or
By a similar calculation, the condition for irrotationality
of a two-dimensional flow
Figure 11-5 Polar coordinates notation.
2( = ou- ov = 0
y oy ax
becomes in polar coordinates
2( = !o(rvo) - ! ov, = 0
r or r o8
Note when v, = 0, rv 0 = constant. It is a free vortex.
Part 2: Some Mathematical Treatments of the Basic Equations
11-4.2.2 Some examples of elementary combinations 11-4.2.4 A source is a flow that moves radially outward
of flow patterns without circulation are, as shown in from a point assumed to be infinitely small (Fig. 11-6).
A sink is a flow radially inward to a point.
Fig. 11-7,
In practice, such a flow is fairly well represented by the
flow through a porous medium toward a well of small
Figure 11-6 Basic flow patterns. diameter, insofar as the vertical component is small, i.e.,
insofar as the curvature of the water table is small.
However, as previously mentioned, its main interest
lies in the fact that complex flow patterns usually en-
countered in engineering practice may be obtained by a
combination of sources, sinks, and other elementary kinds
of flow.
Let Q be the discharge of the source. The components
of velocity at any point are v6 = 0 (for the purpose of
SINK VORTICES
symmetry) and
Q a 1 ol/f
v =~=-=--
r 2nr or r ()()
Stream function and velocity potential function are
given by direct integration. They equal, respectively,
1/1 = _g_ ()
2n
<P=~lnr
SOURCE
Chapter 11 : Flow Pattern, Stream Function, Potential Function
125
(c)
Figure 11-7
(a) (b)
Examples of
combination of basic flow
patterns without
circulation.
(e)
FLOW PAST A CYLINDER
(d)
Equipotential lines, given by = constant, are circles the velocity distribution is governed by the law v9 r =
(r = constant). Streamlines, given by 1/J = constant, are constant = K/2n. It may be seen that when r tends to zero,
straight radial lines (8 = constant). Changing Q to - Q v9 tends to infinity. Such a motion is irrotational.
gives the velocity potential function and stream function As there is no radial flow
of a sink.
It is easy to verify that the velocity potential function _ o _ ~ oi/J _ 0
v,- or -roe-
of a three-dimensional source, where V = Q/4nr 2 is
= - (Q/4nr). In this case, the equipotential surfaces Hence, one obtains
= constant are spheres (r = constant).
v9 = V = -
K 1 o
= -- = - -
ol/f
11-4.2.5 A vortex is a flow in which the streamlines are 2nr r oe or
concentric circles (Fig. 11-9). In a "forced vortex" water
which yields
turns as a monolithic mass, the velocity being proportional
to the distance from the center (Section 2-4.1 ).
The flow under study is a "free vortex." In a free vortex 1/1 =- J!S_
2nr
dr = - K In r
2n
Part 2: Some Mathematical Treatments of the Basic Equations
126
(c)
The flow pattern is very much the same as that of a source 11-4.3 Circulation of Velocity
or a sink, but the streamlines and equipotential lines are -Definition
interchanged. As the flow is irrotational, the Bernoulli Circulation is a mathematical concept on which the
equation may be applied throughout the fluid: theories of wings, airfoils, blades of pumps or turbines,
v2 p* propellers, fans, the Magnus effect (which causes deviation
-. + - = constant of a tennis ball), some sand motions in a flow, etc. are
2g 05
This yields based.
K2 p* Circulation is given by the integral along a closed curve
-8 2 2 + -= = constant S of the tangential velocity component V s
n r g w
It is interesting to note that when r tends to zero, p* /05
tends to - oo. Hence, the presence of vortices in a flow
is a very important cause of cavitation when air is not
admitted into the core from a free surface. Then capillarity It can be demonstrated that r is equal to zero in an
forces take on importance when r -+ 0. irrotational flow.
Chapter 11 : Flow Pattern, Stream Function, Potential Function
Examples are first given; then the conditions for adding 127
flow patterns, velocity potential functions, or stream
functions are analyzed by consideration of the boundary
conditions. Also, some more general considerations of the
methods of calculation in hydraulics are given.
128
a_.---~.::- 6
--~~~~--~~--~~~~--~~~--~~~----~~----5
Figure 11-10
cylinder. Hence the outside flow pattern is considered as the is nil. This result is general. The total force exerted by
flow of a perfect fluid around a cylinder. The stream function uniform stream of a perfect fluid on a submerged body,
for flow around a cylinder is without circulation of velocity, is nil. This is the paradox
of d'Alembert.
K sin(}
1/1 = - + Ur sin (}
r 11-4.4.4 If 4Yt. 2 , ... , ., ... are solutions of V2 = 0,
or any combination = 1 + 2 + + . + is also
1/1 = u(r - ~2 ) sin (} a solution of V2 = 0 and is therefore a possible flow
pattern. A similar rule exists for the stream function 1/1, and
where R = (K/U) 112 and U is the velocity at infinity. It can solutions of V2 !/J = 0. However, boundary conditions
be demonstrated that R is the radius of the cylinder. The have to be satisfied, and will not necessarily allow the simple
potential function is found equal to addition of elementary solutions to yield the final potential
(or stream) function.
= - u(r + ~2 ) cos e 11-4.4.5 Consider the flows presented in Fig. 11-11.
The velocity distribution around the cylinder is In the first case of a flow under pressure, an addition of
solutions characterized by velocities vl and v2 at a given
18 . point does not change the flow pattern because the pattern
V = Voir=R =-;:a(} = 2U Sill(}
does not depend upon the absolute value of the velocity.
and the pressure distribution is In the second case of flow with a free surface, the flow
pattern is changed because the slope of the free surface
(p - Po:J = tpU 2 [1 - 4 sin 2 (}] changes with V. The solutions cannot be added, as they
where p oo is the pressure at infinity. It can be verified that depend upon the absolute value of the velocity. This stems
the net pressure force on the cylinder from the fact that the flow depends upon a non-linear
equation: the momentum equation; or more specifically the
f"/4 Bernoulli equation, in which the elevation of the free surface
F = 4 Jo p cos (} R d(}
is related to the square of the velocity. The first flow pattern
Chapter 11 : Flow Pattern, Stream Function, Potential Function
Hence, the flow pattern and the velocity field on one side, 129
and the pressure distribution and free-surface streamlines
on the other side, cannot be calculated independently by
under pressure may be drawn directly from the fixed bound- 11-5 Reflections on the Importance of
ary, which defines two streamlines. This flow pattern Boundary Conditions
depends only on linear relationships:
11-5.1 New Theoretical Considerations
1. The continuity: au + av = 0 or vz = 0 on the Kinds of Flow
ax ay
From the previous considerations, it is seen that in any
2. The irrotationality: au- av = 0 or V2 t/J 0 kind of flow the method to be used to determine the flow
ay ax =
pattern depends upon the kind of boundary conditions
3. The boundary and upon the assumption of irrotationality. From this
acp = 0 point of view, two major categories of motion may be
condition: an distinguished that are encountered in all methods in hydro-
This boundary condition involves the continuity only. dynamics: analytical, numerical, and graphic methods,
This flow pattern does not depend upon the absolute or methods based on an analogy. The major categories are
value of velocity, only upon its relative value. In a word, the on one hand the irrotational motions under pressure
solution for the flow pattern under pressure within given and slow motion, and on the other hand, the free-surface
boundary is unique. flow and flow with friction force.
Then the pressure distribution may be calculated in a
final, independent step by application of the momentum 11-5.2 Irrotational Motion under
equation in the form of the Bernoulli equation.
Pressure and Slow Motion
In the second case, the flow has a free surface. This free The first category includes all irrotational motions under
surface is unknown and must be calculated. Both the pressure, or motions considered as such, and slow motion
nonlinear momentum equation and the continuity equation in which the quadratic terms are negligible.
must be taken into account.
The boundary condition at the free surface p = constant 11-5.2.1 In the case of flow under pressure, the stream-
involves a force and must be introduced in the momentum lines at the boundary are fully determined since they are
equation to calculate the shape of the free-surface stream- coincident with this boundary. The boundary conditions
lines. In turn, this shape has an effect on the flow pattern. are expressed to satisfy the continuity principle, that is,
Part 2: Some Mathematical Treatments of the Basic Equations
130 that the velocity is tangential to the boundary. The flow boundary conditions as well as from the linear continuity
pattern depends completely upon linear equations only, relationship.
expressing the continuity and the irrotationality principles, Difficulties arise from this nonlinearity and the fact
after which the flow pattern is relatively easily known. that the free surface is unknown. 1 and 2 as defined in
The velocity field gives the relative value of the velocity. Section 11-5.2.2 may also be calculated at a higher order of
The absolute value can be known when the velocity is approximation from the nonlinear free-surface conditions,
determined at one point, either given by a boundary con- which take into account the convective inertia term. How-
dition or calculated by application of the momentum ever the solution , representing the two nonlinear waves,
equation at this boundary. cannot be obtained by simple addition of the new solutions
Finally, the pressure distribution is determined from the 1 and 2 given at a high order of approximation. It must
knowledge of the velocity at any given point by application be calculated from the basic equations. Similar consider-
of the momentum equation. ations prevail in the case of irregular waves traveling at
different velocities in the same direction. There is nonlinear
11-5.2.2 In the case of slow motion, the motion is mathe- interaction.
matically considered as infinitely small, even with a free In conclusion, in order that the velocity potential
surface. Hence, all the quadratic terms may be neglected functions and the stream functions may be added, they must
and the momentum equation becomes linear. The free depend upon a linear and homogeneous equation only. Also,
surface is considered to be known at the beginning and is the boundary conditions should be homogeneous.
denoted by a horizontal line. In that case, various solutions
of flow patterns may be added. 11-5.3.2 Similarly, a friction force (resulting in a rota-
For example, if 1 is the velocity potential function of a tional term different from zero) gives a boundary condition
periodic gravity wave at the first order of approximation, V = 0. Such a boundary condition must also be introduced
that is when the convective inertia term is neglected, and if in the momentum equation. The nonlinearity of the momen-
2 is the potential function of another wave traveling in the tum equation, caused by the convective inertia term, is the
opposite direction, 1 and 2 are determined by a system major cause of difficulty in studying this kind of flow.
of linear equations, as will be seen in Section 16-3. These mathematical difficulties show the importance of
Hence, = 1 + 2 is the potential function of the the irrotational motion under pressure and of the slow
resultant motion. motions in hydrodynamics, even if they only represent very
approximately the natural conditions.
11-5.3 Free-Surface Flow and Flow with
Friction Forces
11-6 Flow Net
The second category of motion includes all motions
with free surfaces or the motions for which the friction
11-6.1 Flow Net Principle
forces have a nonnegligible effect, causing the motion to be The flow net is a family of equipotential lines and a family
rotational. of streamlines representing a complete two-dimensional
flow pattern, as was shown in Fig. 11-4.
11-5.3.1 The free-surface condition involves a force The equalities V = ocpjoS = oljljon for a finite difference,
(p = constant). This force can only be inserted in the take the form V = f'!..1 f'!..S = f'!..ljl If'!..n.
momentum equation, which is an equality of force. First, f'!..ljl is chosen to be a constant in the complete
The flow pattern is now determined from nonlinear velocity field, which means that the discharge f'!..Q between
Chapter 11 : Flow Pattern, Stream Function, Potential Function
L
y
net with a free surface without an experiment. This is even 133
more true when it is a flow between two free surfaces such
as a free-falling jet. Most often the necessary experiments I
I
for determining the free surface are self-sufficient for I
practical engineering purposes. Because a model must be X
built, it can also be used for measuring the pressure distri- Figure 11-15 Notation for conformal mapping.
bution; and the determination of the flow net is then a purely
academic exercise.
It is more accurate to determine the flow pattern by a
numerical scheme and computer. However, the pro- without circulation of velocity. Only the principle is
grammed numerical scheme is based on the same trial- described below.
and-error method. Conformal mapping (Fig. 11-15) is based on the use of
complex numbers and functions of a complex variable.
Two complex planes are used. In one, W = cp + it/1,
11-6.4 Other Methods, Conformal whereas in the other Z = x + iy = reiO. A relationship
Mapping between the two planes is developed such that W = f(Z).
A conformal mapping operation consists of establishing
11-6.4.1 There are a number of methods for drawing a a relationship between each point of a given flow pattern
flow net. All of them are based on the same principles, and in the x, y plane and a point of another flow pattern in the
a similar difficulty is encountered in the determination of a cp, t/1 plane. The first is often the real flow under study;
free surface. The most satisfactory solution is found by the second is often a uniform or a simpler flow pattern.
using a trial-and-error numerical method with a computer. Successive conformal mapping operations may also be
The relaxation method is also based on numerical calculus. done in order to pass step by step from a very complex
An analogy with an electric field is very often used. In this flow pattern to a uniform flow.
case, the analogous equipotential lines between boundaries Conformal mapping can also be used for determining
at different voltages are measured directly. This method free streamlines (Schwartz-Christoffel transform). However,
is used with liquid resistance mesh, wetted earth, etc. its application requires that gravity forces be neglected.
The relaxation method and electrical method may be
easily extended to three-dimensional irrotational flow.
Another analogical method is based on the fact that the 11-6.4.3 For example, consider the transformation
mean motion of laminar flow at constant thickness may be
considered to be irrotational (see Section 2-6.4).
Finally, because the mean motion of laminar flow
through a porous medium is irrotational, it is very easy
to use the analogical method to study any two-dimensional where W = cjJ + it/1 is the equation for a uniform flow in
or three-dimensional patterns. the W plane, i.e., in the system of cp, t/1 axes. This flow is
parallel to the cjJ axis,. and the streamlines defined by
t/J = constant are perpendicular to the t/1 axis.
11-6.4.2 It is beyond the scope of this book to develop The above relationship is the transformation of a flow
the powerful mathematical tool of conformal mapping around a cylinder of radius R into a uniform flow. This
for studying two-dimensional irrotational flows with or can be seen when the real and imaginary parts are separated.
Part 2: Some Mathematical Treatments of the Basic Equations
134 This is easily done by substituting Z = rei 0 into the trans- Flow past a cylinder with 2
formation and using the relationship eiO = cos (} + i sin e. circulation of velocity:
W = U( Z + -R ) - !__
K In Z
Z 2n
= u(reiO + ~2 e-iO)
PROBLEMS
= u(r cos(} + ~2 cos(} + ir sin(} - i ~2 sin(}) 11.1 Draw a square mesh in a two-dimensional bend such
as shown in Fig. 11-16 and calculate the relative pressure
= u(r + ~2 ) cos (} + iU(r - ~2 ) sin (} distribution on both boundaries and along the streamline
starting from point A at the center of the upstream pipe.
Then it is seen that the potential function is 11.2 Give the expression for the Navier-Stokes equations
as a function of the stream function t/J(x,y) in the case of
two-dimensional motion
Figure 11-18
Figure 11-19
3
Chapter 12 12-1 External Forces and Internal
Forces
137
dV dS = m -dV V dt = d -
F dS = m-
(mV 2
)
number of examples in this chapter.
A considerable number of hydraulic problems are
dt dt 2 simplified by the fact that the sum of the internal forces is
The separation between internal and external forces is zero. It is for this reason that the momentum theorem is so
always possible. Hence often used.
The momentum theorem is used to calculate the overall
I:(Fe dS) + I:(F; dS) = d(m;
2
) effects of a mass of fluid, however complex the flow, without
dealing with the fine structure of the flow pattern. However,
However, in spite of the fact that the total sum of the in- to apply the method which consists of considering only the
ternal forces is zero by definition (I:F; = 0), the work of external forces to calculate the change in momentum
these internal forces does not equal zero, i.e., I:(F; dS) =I= 0. requires a perfect knowledge of boundary conditions at the
To illustrate this point, consider a uniform flow in a extremities of the mass of fluid under study. This point is
pipe (see Fig. 12-2). The external forces acting at the wall- illustrated in Section 12-4.
boundary have a total sum that is different from zero,
thereby tending to move the pipe in the direction of the flow. 12-1.4 Momentum Theorem and
But the existing internal forces sum to zero. However, these Navier-Stokes Equation
internal forces do work and this work is the cause of the The momentum equation for a finite mass of fluid can
head loss. The head loss expresses the transformation of be established in several ways as the Bernoulli equation.
energy lost by friction into heat. One could make use of the basic Navier-Stokes equation
Thus, insofar as the energy equality is concerned, internal by integrating all the forces causing motion of an elementary
forces may not be neglected. particle of fluid mass p to the forces involved in the motion
Chapter 12: The Momentum Theorem and Its Applications
A
v
I
I
I
I
I
I
L2:;:::__..,.-r::;;.-r--..- Vn
Part 2: Some Mathematical Treatments of the Basic Equations
dt f{ pvv,. dA + dt f{,pvv,. dA
take the following forms:
2
f{pvv,. dA
Finally, the total change of momentum per unit time with
respect to both time and space is equal to the sum of the
external forces. Therefore
An averaging correction factor (1 + Q(/3) should be applied When the momentum theorem is applied along a hori- I4I
to the term pQV. This correction is neglected in the following zontal axis, the gravity force is zero.
(see Section 10-3.1). The shearing forces, acting at the limit of the considered
The external forces are composed of pressure forces, volume of fluid, are often neglected in the case of short,
gravity forces, and shearing forces. For reasons discussed rapidly varied flows, such as given by a sudden enlargement
below, the latter two forces are often neglected, leaving or a hydraulic jump (see Section 12-3.3.).
only the pressure forces to consider.
The pressure forces on sections A 1 and A 2 may be 12-3.2 Specific Force
considered as consisting of two components The specific force in a cross-section A is the sum of the
momentum flux and the pressure force per unit of weight of
1. The sum of the forces caused by constant pressures fluid, i.e.
Pz and Pb i.e., pzAz- P1A1.
2. Hydrostatic forces applied to the center of gravity of vz p)
(-+-A
the cross sections (Fig. 12-7). In the case of a free g pg
surface two-dimensional flow on a horizontal bottom In order to illustrate previous considerations and to
of depth h, the external forces per unit width are: provide a guide for further applications, some examples are
given with an emphasis on all the necessary assumptions
not usually given in elementary textbooks on hydraulics.
X
0
Part 2: Some Mathematical Treatments of the Basic Equations
--'-------,- J3..f:..Wf!..E_LI NE
1 ERIC PRESSURE-----
This is the hydraulic jump equation. Choosing a second
axis in the vertical direction will give an equality between
the atmospheric pressure force acting on the free surface,
the gravity force, which is equal to the total weight of water GATE
in volume ABCD, and the external force acting vertically
upward on the bottom of the hydraulic jump. There is no
change of momentum in this direction.
Chapter 12: The Momentum Theorem and Its Applications
143
12-3.5 Paradox of Bergeron
Consider a tank on wheels as in Fig. 12-9. On one side
the pressure distribution is hydrostatic, whereas on the
other side the pressure head is transformed into velocity
head. The difference of total forces acting on both sides of the
tank could be obtained by calculating, successively, a flow
net, the pressure distribution on the two sides, and the forces.
Figure 12-10 Paradox of Bergeron:
However, the momentum theorem gives the total value of
A - (B + C) = E - D.
the force directly as
F = pQV = pCcA(2gh) 112
The same result is obtained in the case shown by Fig.
where Cc is the coefficient of contraction. 12-11. The jet acts on the wall of the downstream tank.
Because of this force and the fact that other forces (such The tank does not move. Finally, consider the two tanks as
as gravity and atmospheric pressure) have a horizontal shown in Fig. 12-12 in which it is assumed that the holes
component equal to zero, the tank has a tendency to have the same cross section. One of the holes is closed by
move in the opposite direction of the jet. This is the principle
means of a plane held in place by the jet from the left tank.
of jet propulsion. The area of pressure ABCD equals the area A'B'C'B' or
Now suppose that water is present outside the tank and ABE+ FDC = E'B'C'F'. Considering the forces on the
that the tank is heavier than the buoyant force (Fig. 12-10).
plate we obtain F L = F R when
In this case, the tank does not move. The force caused by
the jet is equal to the force caused by the very complicated pACA2gzt) = pgAz 2
motion inside the tank. This may be considered as another
i.e.
application of Newton's third law-action is equal and
opposite to reaction-and the momentum theorem must be 2Cdz 1 = z 2
applied to the total mass of water. This is what is known as
With cd = 0.60 for an orifice, it is possible for Zz to be
the Paradox of Bergeron. 1.2 z 1 by the simple insertion of a plate. Shaping the hole
144 by the pipe, the total force on the pipe is obtained by calcula-
I
ting the difference of pressure on both sides (inside and out-
I side) of the bend. The projected cross section is A. The force
L_ -tt-------,~-- - -: -+ -. . ,. ~r-tt-- J
is Fe= PeA outside. Inside the bend, the force projected on
A is Fi =piA+ pQV. This is caused by the change of
momentum flux direction in the bend. However, application
of the Bernoulli equation gives
Figure 12-13 Jet reaction compared to intake. Figure 12-14 When the water is expelled, a reaction is exerted.
F=pQV F =O
Chapter 12: The Momentum Theorem and Its Applications
145
12-4 Difficulties in the Application of
the Momentum Theorem
12-4.1 Difficulties in the Case of
(I) (2)
Unsteady Flow
Figure 12-15 Sudden enlargement.
In the case of steady flow, the integral with respect to
volume D(fffv (apVjat) dD = 0) and the momentum theo-
rem becomes simply streamlines have such a curvature that this assumption is
incorrect (see Fig. 12-16). However, it is known by experi-
LFe = JipVY.t dA ment that when the Reynolds number is greater than a
critical value, the flow enters the wider pipe in the form of a
As for external forces, only the boundary conditions for jet. This jet, often unstable, generates by friction some
A appear in this equation because the momentum flux is secondary currents in the corners.
given by a surface integral. Hence its application does not If the velocity is small enough for the convective inertia
require a knowledge of the fine structure of flow within the to be negligible, it is true that the pressure distribution at the
domain D but only on the surface A. cross section (1) is hydrostatic and this usual assumption is
In the case of unsteady flow, the volume integral valid. In fact, the eddies caused by friction induce a centri-
fJJv (apVjat) dD is different from zero and requires a fugal force that causes the pressure force to be slightly
knowledge of the flow patterns within D as a function of greater than the force calculated with the previous assump-
time and space. Thus, the momentum theorem is difficult to tion. However, the assumption is quite valid for practical
apply to unsteady flow and as such, is less frequently used purposes.
except when V(t) can easily be determined.
In the case of steady motion, the momentum theorem 12-4.3 Hydraulic Jump Caused by a
permits the analysis of complex motions. This leads one to Sudden Deepening
think about the precautions that have to be taken in the
This example illustrates a case where it is impossible to
application of this theorem.
calculate external force without experiments.
One difficulty in the application of the momentum
Suppose that a channel has a sudden deepening in order
theorem is in the choice of the boundary and the boundary
to fix the position of a hydraulic jump (Fig. 12-17). The
conditions. It is impossible to calculate the external forces
external force exerted by the vertical wall changes as shown
without having a rough idea of the flow pattern. This often
in the diagram, depending upon the exact location of the
requires experimentation or knowledge of similar previous
experiments. Some examples to illustrate these considera-
tions are given below. Figure 12-16 Sudden enlargement. Laminar flow.
q6
12-4.5 Intake
Consider a free-surface flow as shown in Fig. 12-19.
The rise of the water level in part ABCD may theoretically
be calculated by applying the momentum equation in the
OX direction to the mass of water enclosed in EFBC.
Thus, if one assumes that the flow is perpendicular to the
limit GD:
Figure 12-17 Hydraulic jump on a sudden
deepening. pQV = pg [ (d + !!d) 2 - d2 ]
~
2 2
which gives !!d = V 2 jg and not V 2 j2g, as may be expected
jump. Only systematic experimentation gives the factors
following a superficial analysis.
that influence this phenomenon, and the corresponding
However, the practical result is often closer to V 2 j2g
theory is dictated by the results of this experiment.
than V 2 jg. This is not because the momentum theory is
wrong, but because the boundary conditions are wrong.
12-4.4 Hydraulic Jump on a Slope
The velocity at section GD is not perpendicular to the cross
The study of a hydraulic jump on a slope involves a body section and the momentum theorem should be applied to
force-the force of gravity-which definitely influences the the mass BCDJHGEF in order to include the difference
flow conditions (Fig. 12-18). The gravity force to be consider- in external forces applied at DJ and HG. Unfortunately,
ed is the component of the total gravity force in the direction these external forces cannot be estimated by theory.
of the main flow due to the weight of water included between
the two usual limits where the flow is parallel to the bottom. 12-4.6 Unsteady Flow-Translatory
Hence this force pgAl sin rx is a function of the length of the Wave
hydraulic jump l. This length could be roughly estimated As an example of unsteady flow where the momentum
by experimentation, but it is evident that it cannot be theorem may be used, the case of a translatory wave is
determined with great accuracy. However, the length of the analyzed (Fig. 12-20). The wave is traveling at a velocityW.
hydraulic jump must satisfy the equation
pQ(V2 - V1) = pg .
(2hi - 2h~) + pgAl sm rx Figure 12-19 External forces due to the intake
cannot be determined.
~F,_------~A~----;9
I I
I I
Figure 12-18 Hydraulic jump on a slope --r-~. - - " j --+--... X
(I = length ofjump).
- ,__\_: c
q8 B
Figure 12-21
Total thrust is given by the momentum
theorem. The pressure distribution is given by
the Bernoulli equation and by a flow net.
F= pQV ~o~ a
C A
In contrast, the precise pressure distribution caused by the work equation are valid to study a diverging flow where
above conditions of flow may be calculated by the following the head loss may be neglected, but there cannot be any
method: application of the energy-work equation to a sudden
The flow is assumed to be irrotational and flow net may enlargement without the introduction of another term
be drawn, which gives the velocity distribution. The pressure expressing the head loss, despite the identity of equations
distribution is given by the application of the Bernoulli in their differential form.
equation expressing the conservation of energy. It is evident
that the total thrust may also be deduced by this process of
calculation, by an integration of pressure forces Jp dA, 12-5.3 Mechanics of Manifold Flow
and this could be compared to the result given by the The mechanics of manifold flow is of particular interest
momentum theorem. as an illustration of the previous considerations on the
However, the result given by the momentum theorem, fields of application of the Bernoulli and momentum
which is obtained without any assumptions, is more exact equations. Consider a flow as shown by Fig. 12-22. If the
provided that the boundary conditions are well known. motion is two-dimensional, a first method of analysis
assumes the motion to be irrotational. In such a manner it
is possible to calculate the flow pattern by conformal
12-5.2 Unidimensional Rotational Flow
mapping, although this method is far from being valid
The momentum theorem may be used to analyze a because of the friction forces.
number of phenomena, such as the head loss at a sudden It is also possible to apply the energy equation to the
enlargement or a hydraulic jump, for example, whatever two stream tubes which separate into the main conduit
the complexity of the flow. flow and the flow in the lateral, if a term is included to
Combining the force-momentum equation with the express energy losses for both stream tubes. Unfortunately,
energy-work equation gives the value of the head loss by it is impossible by theory to establish the value of this term.
calculating the difference in total heads: (V 2 /2g) + (pjw) One can assume that head loss for the lateral pipe stream
+ z. Both the force-momentum equation and the energy- tube is that of a bend, whereas for the main flow it is that of
Chapter 12: The Momentum Theorem and Its Applications
149
PROBLEMS
DAM
Figure 12-24
lOOft R = 3.05m
R =lOft GATE
305m
B
where y 1 and Yz are the upstream and downstream water
depths, respectively, F 1 = Vd(gy 1 ) 112 is the Froude number
of the upstream flow, where V1 is the average flow velocity.
151
.~----.....I:>.J>..I
a) ?~~----roM
Figure 12-27
12.9 A jet hits a plane perpendicularly. The discharge of the Figure 12-28
jet is Q = 2 ft 3 jsec (56630 cm 3 jsec) and the particle velocity
is V = 20 ft/sec (609 em/sec). The plane is moving at a
velocity U (U < V) in the direction of the jet. Calculate (1)
the total force exerted by the jet on the plane as a function of
the velocity U; (2) the power of the jet in horsepower; (3) the
power transmitted by the jet to the plane as a function of the
velocity U; (4) the efficiency defined as a ratio of these two
powers when U varies from 0 to V.
Do the same calculations assuming that the plane is
replaced by a bucket as shown on Fig. 12-29. I
Part 2: Some Mathematical Treatments of the Basic Equations
152
Figure 12-29
153
Figure 12-31
FREE STREAM
Chapter 13: The Boundary Layer, Flow in Pipes, Drag, and Added Mass
LAMINAR FLOW boundary layer. The corresponding decrease in flow rate is 155
TURBULENT FLOW
WITH LAMINAR SUBLAYER
L: 0 (U 0 - u)dy
TURBULENT FLOW
PERFECT FLUID
The displacement thickness 5* is the value by which the
wall will have to be shifted in order to give the same total
discharge if the fluid were frictionless. Consequently, 5*
is defined by the equality, as shown in Fig. 13-3,
(j*Uo = L: 0
(U 0 - u) dy
13-1.1.2 The larger the value of the Reynolds number, 13-1.2.3 The momentum flux is reduced in the boundary
the thinner the boundary layer. layer. As a measure of reduction of momentum flux, the
Consequently, at very high Reynolds numbers, the momentum thickness is then defined in a manner similar
average flow motion with respect to time is very close to to that for the displacement thickness. The reduction of the
that of a perfect fluid. This point has been explained momentum flux in the boundary layer is
qualitatively in Section 8-1.1, and is further illustrated in
Fig. 13-2. L:/u(U 0 - u) dy
u.
u.
Figure 13-4
Schematic drawinq of the boundary layer.
Chapter 13: The Boundary Layer, Flow in Pipes, Drag, and Added Mass
incidences, as a consequence of the Bernoulli equation 13-2.1.3 Let us now consider the case of steady flow 157
applied in the second domain, one has (ap*jax) 0. = over a flat plate where ap* ;ax = 0, and let us introduce the
stream function 1/J(x,y) such that u = aljJ ;ay, v = - (aljJ ;ax)
13-2.1.2 As shown in Fig. 13-4, the momentum flux whereby the boundary-layer equation becomes
per unit width through OM is equal to pU6f>. The momen-
tum flux through AB is smaller. However, because u < U 0 aljJ a2 aljJ a2
t/l a3 t/J t/J
------=V-
in the boundary layer, the definition of shows that < f>.
e e ay axay ax ay 2 al
Hence, the difference of momentum is also of order pU6f>.
In contrast, the total friction force per unit width between which is a third-order nonlinear differential equation. From
between the sections 0 M and AB is the above definition it is seem that t/1 is equal to Jb u dy. If
u = U 0 <1>(ry), and y = ry(vxjU 0 ) 112 are substituted in this
equality the result is
or 112
v= - -aljJ = _o) (ryf' - f)
1- (vU
b ""' (~)1/2 = (~)1/2 ax 2 X
pUo Uo
au = a21jJ = u o(u 0)1/2 f"(ry)
This equation is valid only in the case of uniform pressure ay ai vx
distribution along the plate. If it is assumed that the velocity
profiles at all distances x from the leading edge are similar, au= a21jJ =- U 0 11 f"(ry)
ax axay 2x
the velocity profiles u(y) for varying distances x can be
made identical by choosing the proper scale factors for
u(y) and y. The free stream velocity U 0 and the boundary
a31jJ = u
al
o(uvxo) f"'(ry)
layer thickness f> are these scale factors. Hence, one has
the similarity relationship Substituting these into the boundary-layer equation, one
obtains the Blasius equation 2f"' + ff" = 0, and the
u boundary conditions f = f' = 0 at 11 = 0, and f' = 1
- = <l>(ry)
Uo atry=oo.
defining <l>(ry), where 11 = yjf>, i.e.,
13-2.1.4 The general solution of the Blasius equation
U 0 ) 1/ 2
( -
cannot be given in a closed form. However, the solution
IJ = y
vx can be obtained through power-series expansion. The
Part 2: Some Mathematical Treatments of the Basic Equations
power-series expansion near 17 = 0 is assumed to be of f"(O) = 0.332. Therefore, the shear stress at the wall is
u.- u. u.
C LIMIT OF BOUNDARY
----~~~~~--~LA~YEER
Figure 13-6
Diagram for the momentum integral method.
8 ------------~---x
oL-----~--~------------~
I
t +d
Part 2: Some Mathematical Treatments of the Basic Equations
160 through the control surface. This fluid discharge was 13-2.2.5 Now that all the terms have been established,
shown in the previous section to be it is possible to write the momentum integral equation
for the element ABCD. One obtains the momentum integral
by equating the momentum flux (in the x direction) and
the net pressure force to the shear force on the boundary,
that is,
Therefore, the x momentum inflow over the top boundary
DC is
d
To= dx [fb/u(U 0 - u) dy J+ dU0fb/(U 0 -
dx u) dy
13-2.2.4 The pressure forces acting on the limit of the
element ABCD in the OX direction are now considered. Introducing the displacement thickness, b* and the
The pressure forces on AD are pb, and on CB they are momentum thickness (), the momentum integral can be
written
where a0 , at> a 2 are constants that can be determined from Both the boundary-layer thickness and normalized shear 161
boundary conditions. For example stress are close to the exact value obtained in Sections
13-2.1.5 and 13-2.1.6.
u=O at y=O implies a0 = 0
u = U0 at y=b implies a 1b + a2 b2 = U0 13-2.2.7 For flows with pressure gradient, Pohlhausen
suggested using a fourth-order polynomial
du ~ 0
at y=b implies a 1 +2a 2 b=0
dy- u
U = ao + a 1 1J + a2 1J 2 + a 3 1J 3 + a4 1J 4
Therefore
and adding to the boundary conditions used above, the
Uo
a1 = 2-y conditions
o2 u 1 dp dUo
Uo v - = - - = - U0 - - at
a2 = - -p- oi p dx dx
and
and the velocity profile is
at y=b
J
Infinite Flat Plate
f) = s: ~ 0 ( 1- ~ dy = 125 b
13-2.3.1 Because the plate is of infinite length, the
derivatives with respect to x should be zero. That 1s,
and oujox = 0. From the continuity equation
To au I 2vU 0
au+ ov = 0
---;; = v ay y=o b ax ay
Substituting these equalities into momentum integral it follows that ov/oy = 0. Hence, v is identical to zero
equation, one has because it is zero at the boundary. Furthermore, the
15v = b db pressure p* is constant everywhere, because of the infinite
U0 dx
fluid field. Finally, the Navier-Stokes equation becomes
b = ( vx)l/2
5.5-
This equation is linear. A number of exact solutions can
be found as functions of the boundary conditions. If the
Uo
fluid is moving at a speed U 0 (t) and the plate is fixed, the
and the normalized shear stress is equal to boundary conditions are
u(y,t) = c1 0J y/(2vt)l/2
e~~ 212 dl] + c2
1.6
Uo
Chapter 13: The Boundary Layer, Flow in Pipes, Drag, and Added Mass
4
ljl\
Figure 13-8
Velocity distribution near an
3
kt'
3,.
2 ~ v,
v !/ '\ "(
J 1\
kt 'f
~v
oscillating infinite flat plate. 2
I
I
~
""'
kt 'O,OR2,.
kt ' " \
__.... v
v """ ~ ]/
[7 ~
1'-----t--..
0
- 1.0 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1.0
.....!!....
uo
The solution of this equation is the flow has a tendency to become turbulent. The criteria
of the transition from laminar to turbulent are usually
based on the Reynolds number U 0 x/v, in which U 0 is the
free-stream velocity, x is the distance from the edge of the
The velocity profile, u(y,t) has the form of damped harmonic plate, and v is the kinematic viscosity.
oscillation, with an amplitude of U 0 exp [ -(k/2v) 1 12 y]. The value of Reynolds number (or the location x) at
The amplitude decreases exponentially from the plate. which the boundary layer becomes turbulent depends
The velocity profiles for several instants of time are plotted somewhat on the turbulence level of the free stream; it
in Fig. 13-8. ranges from 10 5 to 10 6 . The shear stress acting on the
A similar solution applies in the case where the fluid is boundary is much larger in the turbulent boundary layer
moving at a velocity u = U 0 cos kt, and the plate is fixed. than in the laminar boundary layer; therefore, the deter-
Such solution is of particular interest for studying the mination of the location of this transition is not only of
motion in the boundary layer of a periodic gravity wave theoretical interest but has practical uses.
and the wave damping by bottom friction. After transition, the main part ofthe flow in the boundary
layer is turbulent. However, immediately adjacent to the
wall, the turbulent fluctuations are suppressed by the
13-3 Turbulent Boundary Layer presence of the wall. The flow field in this region can be
divided into three domains-the laminar sublayer, the
13-3.1 General Description turbulent boundary layer, and the free stream (see Fig. 13-9).
13-3.1.1 For a laminar boundary layer, it has been seen If the boundary is rough, the laminar sublayer may be
that its thickness increases with the distance x from the destroyed by the presence of the roughness elements.
edge of the plate. As this boundary-layer thickness increases, A detailed discussion of this is given in Section 13-4.3.
y
164
--r
FREE STREAM
T
u ( y)
TURBULENT BOUNDARY
y LAYER
!
TRANSITION
/(
o) SMOOTH WALL
Figure 13-9
Turbulent velocity distributions
t-
FREE STREAM
near the wall.
T LAYER
BOUNDARY
Figure 13-10
Schematic representation of the laminar and turbulent
boundary layer.
13-3.1.2 The velocity distribution of a laminar sublayer by the effect of turbulent shear stress which results in a
is determined by the viscous force, that is, T = pv(dujdy). logarithmic velocity distribution, as discussed in Section
Because this layer is very thin, it is reasonable to assume 8-3.5.
that T is constant within this layer and equal to the shear
stress at the wall To. After integration, the equation becomes 13-3.1.4 The effect of the boundary shear stress in the
case of free stream flow is small. Therefore, the flow field
Toy
u= -- can be determined by considering that the flow is non-
p v viscous.
If, by definition, However, there are no sharp boundaries between each
region, and the concept of each domain is to some extent
u* = (~ y/2 qualitative. A schematic drawing of the three domains
of flow and the flow pattern before the formation of the
then one finally obtains turbulent boundary layer is given in Fig. 13-10.
166 layer, it is rather difficult to assume a velocity profile the boundary layer starts to become turbulent, to a given
with sufficient accuracy because of the complicated flow point x measured from the leading edge of the plate.
field. Therefore, some experimental results have to be used The result is
as a base in order to get a good prediction of the variations
of thickness of the boundary layer and the resistance with
distance. Based on the experimental results the velocity
D514 - D614 = 0.29(;J 114
(x - x 0 )
profile can be represented by If one assumes that the boundary layer becomes tur-
8.74(u*y) 117
bulent at the edge of the plate, the initial value above can
!I_= be replaced by D = 0 at x = 0. The above equation gives
u* v directly
where u* = (T 0 /p) 112
is the shear velocity. Because the
u )-1;5
velocity at y = Dis equal to U 0 , then D = 0.37(-;- x 415
R = {To dx = pU68(/)
From the definition of the momentum thickness, one has = / 2 pU6D = 0.036 pU6 (-;-
u )-1;5 1
415
successively
and the resistance coefficient C1 is
R (Uol)-115
cf = 1-----z/ =
zPUo
0.072 -
v
In the range of Reynolds number 5 x 10 5 < Re < 10 7 , the
last equation gives very good agreement with experimental
On substituting momentum thickness (} and the wall shear results.
stress To into the momentum integral equation, The value of the resistance coefficient for the turbulent
boundary layer as well as the laminar boundary layer are
To d(}
plotted in Fig. 13-11.
U0 = constant
pU6 dx
one obtains 13-4 Flow in Pipes
0.0225(-v-)
U D 0
114
= 272 db
dx
13-4.1 Steady Laminar Flow in Pipes
Laminar flow in pipes rarely occurs in practice. However,
This is the differential equation for D. This equation is a rather full discussion is given because the simple and
integrated from the initial value D = D0 at x = x 0 , where rational analysis is of some help in the understanding of
Chapter 13: The Boundary Layer, Flow in Pipes, Drag, and Added Mass
10
-2
After integration, the velocity profile becomes
-- - -- Lip r 2
u =--+c
-
t- TURBULENT 2/f.l 2
~ TR~
~ - --- ----- t- The constant c can be determined from the boundary
condition u = 0 at r = r 0 . Therefore, the velocity distri-
'-
.......... bution in the pipe is
t'- ~INiR
-3
10 Lip 2 2
10 6 106 6 107
u = - (r 0 - r)
R = Uo/ 4/f.l
e 1/
which has the form of a symmetrical paraboloid. By integra-
Figure 13-11 Resistance on a smooth fiat plate.
tion, the average velocity V
Lip f = 64J.1 = 64
du = 21 f.1 r dr VDp R.
-- -
LAMINAR
bution near the wall is based on the assumption that the FLOW
velocity u at a distance y from the wall depends on the
tangential stress r 0 and on the viscosity J1 and density p. SMALL Re..... '\
-+-+-
Therefore, one may write in the most general case that
F(r 0 ,u,y,p,p) = 0 (It is understood that u means the average
velocity with respect to time and should actually be written
ii. The bar will be omitted in the following sections for the ---
sake of simplicity.) Based on the dimensional analysis,
one obtains the dimensionless form a) SMOOTH WALL
u = [('oY/P)l/2]
(ro/P) 112 f v
v v
30 !69
TURBULENT
/
20
v
//
Figure 13-14
Universal velocity distribution law for smooth pipes. 10
21!' l-/
TRA'SOTION
MINAR
I
0
2 3 4 5
10 10 10 10 10
However, in the range u* yjv between 5 to 30, 'Yhere both After integration, one obtains
turbulent and viscous effects are of equal importance, the
velocity profile deviates from both of the above equations.
Mathematical analysis fails to give correct prediction. This
region is usually called a buffer region or a transition region.
On the other hand, it has been seen that the Darcy-
Weisbach equation gives the value of the head loss I!.H
13-4.2.2 Usually the thickness of the laminar sublayer as function of the friction coefficient f and the average
and the layer of transition is very small in comparison with velocity U as follows:
the size of the pipe. Therefore, in computing the average
velocity, one could just use the logarithmic velocity distri- I U2
bution without introducing any significant error. The I!.H =f--
D 2g
average velocity may then be obtained by substituting
where I and D are the length and diameter of the pipe,
u yu*
~ = 5.75log 10 - + 5.5 respectively. Also, pg 11H(nD 2 /4) is the difference of pressure
u* v forces acting on two cross sections separated by a distance
I. This force is balanced by the shearing force To nDI.
into the following equation: Equating these two forces yields
yields where ks is the sand size. (If the sand is not closely packed
(8)f
170
or nonuniform, one should also take into account the
-U =
1 2
- 1 = u r
5.75 log 10 ____>i:___Q + 1.75 concentration distribution and shape of the roughness
u* v
elements.) In the case in which the wall is not completely
(!)
Since
rough, then an additional dimensionless parameter k 5 u*jv
u*r-0 -
- _ 1
--
112
UD
- should also be included. Therefore, the general function
v 2 8 v should read
The above equation can be further written as
/
9
wall. Some typical velocity profiles obtained in rough
pipes are presented in Fig. 13-13b. Following the same 8
procedure for the flow in smooth pipes, one could establish
7
a dimensionless functional relationship for the closely SMOOTH- - TRANSITION COMPLETELY ROUGH -
packed uniform sand roughness elements in completely
rough regime:
0.8
I
1.4 2.0
I
2.6 3.2
log
Chapter 13: The Boundary Layer, Flow in Pipes, Drag and Added Mass
In this regime, the size of roughness is so small it of smooth pipe. The final form of the resistance equation
is covered by the laminar sublayer. reads
2. Transition regime:
1 r0
5 S ksfl* S 70
! 112 = 2.0 log 10 ks + 1.74
v
Experiments were performed first by Nikuradse, who
Some of the roughness elements extend outside the used closely packed sand grain roughness elements and
laminar sublayer and contribute some resistance obtained the resistance diagram shown in Fig. 13-16.
through form drag. The velocity distribution and resistance formula dis-
3. Completely rough regime: cussed are based on the closely packed sand grain roughness
used by Nikuradse. In this case, ks is the actual sand size.
ksfl* > 70 However, if a different type of sand is used or sand particles
v
are not packed closely, the resistance offered to the flow
All the roughness elements are exposed outside the will be different. Therefore, sand size alone is not enough
laminar sublayer or one may say that the laminar to describe the velocity distribution and resistance.
sublayer has been destroyed completely by the
roughness elements. The turbulent action extends
all the way to the rough wall. Further increase of 13-5 Drag on Immersed Bodies
shear and the Reynolds number does not bring any
13-5.1 Draq on a Body in Steady Flow
change of flow patterns. Therefore, B remains
independent of shear Reynolds number. 13-5.1.1 It has been seen that the total force exerted by a
current on a cylinder is zero (see Section 11-4.4.3). This
13-4.3.2 The resistance coefficient in the completely result is general, for the case of a perfect fluid, i.e., the total
rough regime can be evaluated the same way as in the case force exerted on a body by a perfect fluid without circulation
of velocity is always zero. This is called the paradox of
D' Alembert.
Figure 13-16 Resistance formula for pipes roughed with
Nikuradse's sand roughness.
For the case in which a circulation is introduced to the
fluid (see Sections 11-4.2 and 11-4.3) a force perpendicular
to the incident velocity is exerted on the body. It can be
lOr---------~----------~---------,
'o
-=15 demonstrated that this force is proportional to the velocity
ks
V of the fluid and the strength of the circulation. It is this
force that causes the lift of an aerofoil.
1001
The problem now under study is that of a real fluid in
which case a boundary layer develops along the body and
induces a drag. This drag is caused by the shearing force
acting on the body and to the wake. This leads us to discuss
the problem of boundary-layer separation.
1.0 L------------L.----------..l.---------__J
13-5.1.2 The flow field near a flat plate in parallel flow
103 10 4 10~ 106 and at zero incidence is quite simple because the pressure
R
e :~
v in the entire flow field remains constant. In the case of flow
Part 2: Some Mathematical Treatments of the Basic Equations
LIMIT OF BOUNDARY
Figure 13-17
Boundary-layer separation.
about a blunt body such as a cylinder, the pressure along 13-5.1.3 In principle, the total drag exerted on a sphere
the surface of the body, which is impressed on the boundary moving with constant velocity in the infinite flow field
from the external flow, is not constant. As shown in Fig. is the sum of the friction drag (or shear drag) and the form
13-17 the fluid particles are accelerated from A to B and drag. If the velocity is low enough (R. < 1), the inertia terms
decelerated from B to C. Hence, the pressure decreases in the Navier-Stokes equations may be neglected. The
from A to Band then increases from B to C. This can be drag can be obtained analytically and is given by Stokes'
seen by application of the Bernoulli equation. Since the law
fluid is viscous, a certain amount of kinetic energy is lost F = 3npvVD
by the friction within the thin boundary layer as the fluid
particles move along the boundary. The remaining energy where V is the relative velocity of the body with respect to
may be too small to overcome the increasing pressure the water and D the diameter of the sphere. The drag
toward the point C. As a result of this, the fluid particles coefficient Cv, which is defined from the equation
being influenced by the external pressure may move in the CvApV 2
reverse direction and cause flow separation behind the F = --,-2- -
body at point S. The flow field behind the separation is
very irregular and is characterized by large turbulent is then equal to 24/R., where A is the cross-sectional area
eddies. This region is usually called turbulent wake, of the sphere, R. is the Reynolds number VDj v.
although the wake may also be laminar when the Reynolds As the Reynolds number R. increases, the flow separates
number is small ( < 40 in the case of a cylinder). from the surface of the sphere, beginning at the rear stagna-
Because of the existence of the wake, the flow field tion point, where the adverse pressure gradient is greatest.
changes radically as compared with that in frictionless As the flow separates from the boundary, the form drag,
flow. The main flow which separates from either side of which is a function of the area of separation and the square
the boundary does not meet right behind the body. It leaves of velocity, becomes important. The drag coefficient C v
a "dead zone" in which the pressure remains close to its will deviate from the line C v = 24/R., and start to level off.
value at the separation point, which is always less than the Figure 13-18 indicates the variation of the drag coefficient
pressure at the forward stagnation point. Therefore, a C v with Reynolds number R.
large net force will act on the body resulting from the As the Reynolds number reaches 2 x 10 3 , the drag
pressure difference. This force is called form drag. coefficient becomes almost constant. However, in the
Chapter 13: The Boundary Layer, Flow in Pipes, Drag, and Added Mass
400
173
""~
100
Figure 13-18
Drag coefficient for sphere.
10
STOKES " ~
THEORY~
~
-
I
I~
~
range of Reynolds number 2 x 10 5 to 3 x lOS, the drag 5000, one can see a regular pattern of vortices which move
coefficient is suddenly reduced. The reason for this lies alternately clockwise and counterclockwise downstream,
in the transition of boundary layer from laminar to tur- as shown in Fig. 13-21. This is known as the Karman
bulent. This transition brings a violent mixing in the vortex street. The vortex street moves with a velocity V.,,
boundary layer. As a result, the fluid particles near the which is somewhat smaller than the free stream velocity
boundary gain additional kinetic energy, which enables U 0 Von Karman found that the vortex street is unstable
them to withstand the adverse pressure gradient better except at the spacing h/1 = 0.281, and that the drag
and to move the separation point somewhat downstream, experienced by the cylinder depends on the width of the
as illustrated in Fig. 13-19. This results in a sudden de- vortices hand on the velocity ratio V.,/U 0 :
creasing of the drag coefficient near the Reynolds number
3 x 10 5 as shown in Fig. 13-18. As the transition that occurs
depends on the roughness of the sphere, and also slightly F = pU 02 h[ v., 1.12( Uo
2.83 Uo- v., ) 2
]
IIi:!pV (x,y,z,t) da
Figure 13-19 Changing location of separation and
pressure distribution as a result of 2
boundary-layer transition.
100
175
10 ~~
Figure 13-20
Drag coefficient for circular cylinder
r-----1----
\_
0.1
-I I 2 3 4
10 10 10 10 10
R = VD
e II
stop it. It is seen that this work also includes the work also be in general a function of time. However, in the case
required to move the fluid around it: J:M'U 2 Once this of a perfect fluid, V(x,y,z,t)/U is independent of U, but
work is produced, the body will continue to travel in a depends upon the flow pattern only. If one refers V to a
perfect fluid at a constant velocity U. coordinate system moving at velocity U this ratio is also
independent of time. Hence, the integral of the coefficient
13-5.2.3 It is pointed out that because V decreases with V(x,y,z,t)/U is independent of the value of U and the time
R- 3 (or R- 2 in the two-dimensional case), (V/U) 2 varies as well. In a word, M' is a constant associated with the body
with distance as R- 6 (or R- 4 ), whereas the integral of dCJ and the specific mass of the fluid only.
varies as R 3 (or R 2 ). Consequently, the integral for M'
has a finite value. 13-5.2.4 Let us now consider the total force to move
It is seen also that in the general case M' is a function the body. It is equal to the sum of the inertia of this body
of the absolute value of U and consequently of the Reynolds
number U Djv and other empirical parameters charac-
Figure 13-22 Relationship between the Strouhal
terizing the flow (such as UT/D for periodic motion where D number and the Reynolds number.
is a typical dimension of the body). Consequently, M' will
a:: 022
w
ID
::< 0.20 /
Figure 13-21 Von Karman vortex street. ::>
z
...J 0.18
<( I
/
5 0.16
h
0
a::
0.14 I
I
~
CJ)
T CJ) 0.12
I 4
10 102 103 10
R~ REYNOLDS NUMBER
Part 2: Some Mathematical Treatments of the Basic Equations
q6 itself and the inertia of the fluid surrounding it, i.e., should be determined from the force exerted by the fluid
on the body or vice versa, i.e.
JJ'foo dV
fi
dU
F = M dt + P Jlim dt da
F' = p cos eds
which can still be written as
e
where p is the pressure around the body, cos is the angle
F = (M + M')dU of the perpendicular to ds with the main direction of the
dt motion, and s the area of the body. (Fig. 13-23). Given
V (or), p can be determined by application ofthe Bernoulli
where
equation. In general, the integral of p V2 /2 being zero
M' = p(djdt) JJJ!im
V da (paradox of D'Alembert), the integral of p oc/Jjot only is
dU/dt significant, so that finally, replacing p by - pocjJjot
It is not evident a priori that the two definitions for M' M' = - Hp(oc/Jfot) cos e ds
are identical. As a matter of fact the integral dUjdt
Of course, the equality force-momentum can also be
obtained by differentiating the equality work-energy as
follows
may diverge as the distance from the body tends to infinity. !!__ (W) = !!__ [-!-(M + M')U 2 ]
Hence, in the case of moving body the force F' = M' dUjdt dt dt 2
POSITIVE
PRESSURE
Figure 13-23
An example of distribution ofpressure force
a)VELOCI T Y on an accelerating body in still water.
PRESSURE
c l TOTAL PRESSURE
b) ACCELERATION PRESSURE
Chapter 13: The Boundary Layer, Flow in Pipes, Drag, and Added Mass
Since dL/dt = U, the equality force-momentum is obtained. where U(t) is the velocity of the body. The total kinetic
Still, this operation is done by assuming M' constant and energy of the fluid per unit length of the cylinder is then
dFjdt = 0. Actually, a more rigorous demonstration will successively
not require this limitation.
In conclusion, for all practical purposes, the added
mass is determined by calculating the integral
T = f0
2" foo
r=R
1 R4U2
p -2 -
r
4- r dr de
M' = p JJJ(~)2 da
or in the case of an irrotational flow = tpnR2U2
, Hp(a<Pfat) cos e ds It is seen that the added mass isM' = pnR 2, i.e., the mass
M =- dU/dt of a cylinder of radius R having the same density as the
fluid. It is also seen that the total force to move the body is
Then this value of M' will be used for determining the force
M'(dU jdt). The shortcomings of these simple demon- F = (M + M') dd~
strations are not discussed in this book. It is only pointed
out that one of the limits of the integral, namely the body I.e.,
limit, could be time dependent.
dV
F = (pb + p)nRz dt
13-5.2.5 The case of a moving circular cylinder of radius
R is given here as an example. The velocity potential for a where Ph is the density of the body. It can be verified that
cylinder moving through a fluid at rest is given by super- F' = M'(dV/dt) is the total force exerted by the fluid on the
imposing upon the steady state pattern of a flow around a body as the sum of all the pressure forces in the direction
cylinder, a uniform velocity U (see Section 11-4.4.3) i.e., of the motion:
q8 Since the integral of the quadratic term is zero, the total It is seen that the pressure component due to local inertia
force acting on the cylinder is is in this case twice the value of the pressure component
in the case of a moving cylinder.
F' = 5 pR
0
2" dU
2 -
dt
cos 2 8 d8 =
dU
pnR 2 -
dt
Inserting this value in the previous integral yields
I.e., F = 2 nR 2 dV = 2M' dV
p dT dT
M' = pnR 2
13-5.2.8 In the case of a real fluid, this inertial force still
13-5.2.6 Let us now consider the case of a fixed body exists, but because of viscosity, separation, and wake,
subjected to an unsteady current. The total force exerted there is also a superimposed quadratic force. The following
by water on the body is still empirical formula is often proposed
13-5.2.10 In the case in which both a body moves with The first term results from the added mass, as defined in 179
a time dependent velocity V(t), in a time dependent fluid Section 13-5.2 (vide supra). The second term is the result of
flow, which has a velocity U(t) at a distance from the body, the free surface variation and is equal to
the total force to be exerted to move a body of mass M is
nD 2
pCMus4Ws
F = M dV + M' !!__ (V - U)
dt dt
where CM is a dimensionless coefficient, us and ws are the
13-5.3 The Case of Time-Dependent free surface horizontal and vertical particle velocity
Mass components at the pile location (ws = aiJ;at, l](t) being
13-5.3.1 In the case of (for example) an underwater the free surface elevation), and D is the pile diameter.
missile popping up vertically from the tube of a submarine, This force is often neglected. Actually, this neglect intro-
and subjected to a steady current U, an inertial force duces a large error on the calculation of wave force and
acting in the direction of the current is added to the drag torque, in the case of piles subjected to large-amplitude
force. This added force is due to the variation of mass of shallow water waves.
the body subjected to the cross-flow, i.e., the momentum
equation is now written
PROBLEMS
F = dmU = Udm
dt dt
13.1 The dissipation energy thickness ()** of a boundary
where dmjdt is proportional to the rate of volume of missile layer is defined by the equation:
exposed to the fluid flow, i.e., the product
J AdV~
nD 2 pU6 (5** = p Ln u(U6 - u2 ) dy
4 v
u: = ao +at(~)+ a2(~Y
as well as the stream function
Determine the constants a0 , at,
a 2 by use of appro-
priate boundary conditions. Substitute the determined
velocity profile into the momentum integral equation
1. Find the ordinary differential equation for the stream and derive a differential equation for b(x).
function and the boundary conditions. 3. Try a solution of the form c5 = f3xn. Determine the
2. Obtain the velocity distribution. values of f3 and n from the integral equation.
Chapter 13: The Boundary Layer, Flow in Pipes, Drag, and Added Mass
= u(r + ~2 ) cos e
p =p = CONSTANT
0 where U is the velocity at infinity. Give the pressure distri-
bution around the cylinder in the case where
1. U = constant = U 0
2. U = U 0 sin kt
Determine the total force acting on a cylinder by integration
of the pressure.
Open-Channel Formula
14-1.1.1 Three kinds of approaches are used to obtain
section A of an element of fluid A .1x are in equilibrium. equation for Vis obtained r83
The OZ components of the pressure force and gravity also
balance independently of the flow velocity. Thus, the only
significant forces which remain are the gravity component
V = (y) 112
(RH sin 8) 112
in the 0 X direction and the shearing forces: The Chezy coefficient of dimension (LT- 2 ) 1 12 is Ch =
J:
(g/!) 1 12 ; thus f = g/C~. The slope S is generally small, so
pgA .1x sin (} = .1x < dP that S = tan(} ~ sin(} and V = Ch(RHS) 112 This is the
Chezy formula.
where P is the "wetted perimeter" (i.e., the length of the The discharge Qn = VA is then
perimeter of the cross section A which is underwater) and r Qn = ACh(RHS) 112 =KS 112
is the shearing stress per unit area.
where K = ACh(RH) 112 is the conveyance of the channel
and depends upon the geometry of the cross section of the
14-1.1.3 In general, r is not constant, except around a
channel and the water depth only. Qn is the normal discharge,
circular pipe due to symmetry. However, owing to secon-
which is defined as a function of water depth for a given
dary currents, the variation of r can often be considered as
channel.
negligible (see Section 8-2.3).
The previous equation is then written 14-1.15 The Chezy coefficient and f can only be deter-
mined by experiment. It is found that in the case of turbulent
A
pgRH sin (} = r where RH=- flow over a rough bottom
p
- 1.486 1/6
Ch ---RH
Note that the "hydraulic radius" RH has the dimension of a n
length. It is easily verified that in the case of a rectangular where R H is in feet and n is the Manning coefficient; n is
section of depth h and width l given as a function of relative roughness and in practice
lh varies between 0.01 and 0.05. Inserting this expression in the
RH = l + 2h Chezy formula gives the Manning formula
and RH tends to h when l --+ oo; i.e., in the case of a large V = 1.486 Ri/3S1/2
n
river. In the case of a circular pipe of radius R
and the conveyance
nR 2 R
RH=--=- K = 1.486 ARi/3
2nR 2 n
14-1.1.4 In the case of a river or a channel, the Reynolds 14-1.2 Normal Depth and Transitional
number is generally large, and the flow is fully turbulent. Depth
The shearing stress can then be assumed to be related to the
14-1.2.1 The normal depth hn is defined as the distance
average velocity V by a quadratic function such as
between the lowest part of the channel and the free surface
r = pfV2 of a uniform flow. It is determined by the equality
where f is a dimensionless friction factor. When this is
combined with the equation pgRH sin(} = r the following
$~2 = K(hn)
Part 2: Some Mathematical Treatments of the Basic Equations
where K(hn) is the function that characterizes the conveyance this formula because Vis average velocity. For the sake of
of the channel. In the case of a wide rectangular channel, simplicity, oc will be neglected in the following discussion.
since RH = hn, the normal depth hn is It is seen that the specific energy is the sum of the kinetic
qn
( chsl/2
)2/3 energy (V 2 /2g) and the potential energy per unit of weight
with respect to the bottom of the channel h (but not with
hn =
respect to a horizontal datum) (see Fig. 14-2).
where qn is the normal discharge per unit width. It is seen The specific force is
14-2.1 Definition of the Specific Energy Figure 14-2 Notation for an open channel.
and Specific Force
The quantity
_ (1 + oc) V 2 h _ (1 + oc)Q 2 h
E- 2g + - 2gA 2 +
where V is the average velocity, h the maximum water
depth, Q the discharge, and A the cross section, is called the
specific energy; oc is a positive coefficient which appears in
Chapter 14: Open-Channel Hydraulics
h h
I q INCREASES
I
/,
Ec E = :-i!_ +h !=
v2
9fl+2
h2
2g
Figure 14-3 Specific energy and specific force.
14-2.2 Definition of the Critical Depth 14-2.2.3 In the case of a complex cross section A(h) the
critical depth is defined by
14-2.2.1 Let us consider the two functions E(q,h) and
l(q,h) in the case of a rectangular channel. It is seen that for aE 2 Q2 aA
a given value q, E(h), and J(h) vary as shown on Fig. 14-3a,b. oh = - A 3 29 oh = 0
1
In particular, the minimum values for E and I given by Since oAjoh = l, l(h) being the width of the river at the
oEjoh = 0 and oljoh = 0 are obtained for the same value free surface, one has
of h = h0 when Qz A3
g
Because A and l are functions of h only, it is then possible to
establish the curve he as a function of the critical discharge
i.e., Qe for any kind of cross section (see Fig. 14-4). It is easily
verified that in the case of a rectangular cross section, the
he = ( qg2)1/3 or V = (ghe)l/2 previous value he is found.
in which case Figure 14-4 Variation of the critical depth with discharge.
r86 i.e.,
14-3 Tranquil Flow and Rapid Flow g A
14-3.1 Definition
se = c; RHt
and in the case of a large rectangular channel Sc = gjC~ = f
There exist two possible values, h 1 and h 2 , for the same
When S > So one has hn < he corresponding to a value of
values of E and q, defined by
the Froude number V/(gh) 1 12 < 1: the flow is subcritical or
Vi v~ tranquil. When S < So one has hn > he corresponding to a
E = h1 + - = h2 + - value of the Froude number Vj(gh) 112 > 1, the flow is
2g 2g
supercritical or rapid.
and
14-3.3 Free Surface Disturbance and
Flow Control
such that Since any disturbance travels at an approximate speed
(gh) 112 , the occurrence of V > (gh) 112 means no perturba-
tion can travel in an upstream direction. Also, when the
flow characteristics change from a tranquil upstream flow
to a rapid downstream flow such as at the top section of a
By inserting the value for he = (q 2 jg) 113 , one obtains
weir, the discharge is controlled at this critical cross section.
2hih~ = h3
h2 + hl e
14-4 Gradually Varied Flow
The larger depth, say h 1, corresponds to a "subcritical" or
"tranquil flow" in which case V1 < (gh) 112 The smaller 14-4.1 Basic Equations
one, say h2 , corresponds to a "supercritical" or "rapid 14-4.1.1 The generalized Bernoulli equation with friction
flow" in which case V2 > (gh) 112 can be applied to a gently sloped free surface flow as
follows:
14-3.2 Critical Slope
It has been seen that the normal discharge is a function
~ (v2 + !!._ + z) =
dx 2g pg
of the conveyance and the bottom slope: Qn = K(S) 112 ,
whereas the critical discharge is a function of the cross I.e.,
section and the width of the channel only Qe = (gA 3 j/) 112
There is a slope, the" critical slope," for which the normal
discharge, i.e., the discharge of a uniform flow, is equal to
the critical discharge, and consequently the normal depth where
hn is equal to the critical depth he. Q2 Q2
Consequently, the critical slope is defined by --=c"'~R-=-H-A"z = -K-2 = Sf
S1 is the slope of the energy line or head loss. Then, insert
p = Pa (atmospheric pressure) as a constant and call z the
Chapter 14: Open-Channel Hydraulics
elevation of the free surface with respect to a horizontal one has finally,
datum. Since dh 1 - (Q 2 jK 2 S)
dz dh dx = S 1 - (Q 2 ljgA 3 )
-S+-
dx dx which is the fundamental equation for gradually varied
(see Fig. 14-5) the result is flow. Considering that
and
Figure 14-5
Notation for gradually varied flow .
DATUM
: ''':"!2~0.',i;c(T'''~;;:y~.;~'<;\l'h"'\,~;~i'ifu~t%;'i<;~;;c),~:r-<\(9
tJ. X
Part 2: Some Mathematical Treatments of the Basic Equations
I'
_j
~
0z u he
N .c;
ii:A ' '
0 c
J: .c;
w
Figure 14-6
Different kinds of backwater curves.
<1.
0
_j
.-MI
l -
~-
M2
-
en
hr r
w u
n hn
J
_j.c;
he he .......r~c hn
li;:A .L M3
w c
(!).C:
' ' I
I .. L
~- '
w
<1.
0
_j
en
_j
<( u
u.c:
i= II
fj ~c
w
<1.
0
_j
en
u
<1. .c;
w
wv
1- c
en .c:
channel of width unity. The discharge is q. The momentum where h 1 and h2 are the "conjugate" depths. For every
theorem applied to the hydraulic jump is then (see Section upstream value h 1 < h"' there corresponds a value h2 > he.
12-3.3): The backwater curve h'1(x) for rapid flow is calculated by
application of the Bernoulli equation from the upstream
conditions. The backwater curve for tranquil flow h2(x)
is also determined by application of the Bernoulli equation
Part 2: Some Mathematical Treatments of the Basic Equations
--I
CONJUGATE
OEPTH
hI
--
.........
\ TRANQUIL FLOW
Figure 14-7
/ Determination of the location of
--
an hydraulic jump.
HAP/0 FLOW .. ~
~~ _ pg + Pv 2 [h : z ~:~ J
Chapter 14: Open-Channel Hydraulics
which, when integrated, is where S is the bottom slope, Q the discharge, Q" the normal
14.3 It is possible to define a channel that has a constant 14.8 Establish under which conditions there are two normal
hydraulic radius Ra whatever the water depth h? depths and two critical depths in a circular closed conduit.
Determine all the possible backwater curves that may exist
14.4 Demonstrate that the free surface slope of a steady in such a conduit.
gradually varied flow in an open channel is equal to the
sum of the energy slope and the slope due to the velocity 14.9 Establish the required condition for establishing a
change d(V 2 j2g)jdx. hydraulic jump on a sudden bottom drop corresponding
to various assumptions on the exact location of the jump
14.5 Demonstrate that with respect to the sudden bottom drop.
Figure 14-8
DOWNS TREAM
sz= LEVEL
Sabersky, R. H., and Acosta, S. A., Fluid FlQw-A First Course in 193
Fluid Mechanics. Macmillan, New York, 1964.
Schlichting, H., Boundary Layer Theory. McGraw-Hill, New York,
1960.
Streeter, V. L., editor, Handbook of Fluid Mechanics, McGraw-Hill,
New York, 1961.
Taylor, G. I., Statistical theory of turbulence. Proc. Roy. Soc. Lond. A,
151, 1935.
Townsend, A. A., The Structure of Turbulent Shear Flow. Cambridge
University Press, London, 1956.
Von Karman, T., Collected Works. Butterworth Scientific Publica-
tions, London, 1956.
Von Karm{m, T., and Howarth, L., On the statistical theory of
isotropic turbulence. Proc. Roy. Soc. Lond., 164, 1937.
PART THREE
transverse components still exist although they are difficult surface elevation traveling along the OX axis at a velocity C.
to detect. The characteristics of a progressive wave remain identical
for an observer traveling at the same speed and in the same
15-1.1.2 It has to be expected that, due to this inherent direction as the wave (Fig. 15-2). In the case where 11 can
complexity, a simple introduction to the problem of water be expressed as a function of (x - Ct) instead of (x,t), a
waves is a difficult, if not impossible, task. Hence, this "steady state" profile is obtained. I'J(X- Ct) is the general
chapter should rather be regarded as a guide for the expression for a steady progressive wave traveling in the
following chapters and for continuing further study beyond positive 0 X direction at a constant velocity C. In the case
the scope of this book. where the progressive wave is moving in the opposite
The full assimilation of the subject leading to a clear-cut direction, its mathematical form is expressed as a function
understanding of this chapter can only come after a compre- of (x + Ct). It is pointed out that the definition of wave
hensive study of each existing theory within or beyond the velocity C for a nonsteady profile is not strictly valid, since
scope of the present book. With this in mind, the following each "wave element" travels at its own speed, resulting in
classification is proposed. A physical classification is given wave deformation.
first, then the different mathematical approaches and their
limits of validity are introduced. Finally, the traditional two 15-1.2.2 The simplest case of a progressive wave is the
great families of water waves are presented. wave which is defined by a sine or cosine curve such as
WAVE DIRECTION
Figure 15-1
Oscillatory wave.
Chapter 15: An Introduction to Water Waves
199
15-1.2.3 A standing or stationary wave is characterized A standing wave generated by an incident wind wave is
by the fact that it can be mathematically described by a called clapotis. In relatively shallow water (d/L < 0.05) it
product of two independent functions of time and distance, is called a seiche. A seiche is a standing oscillation of long
such as period encountered in lakes and harbor basins. The ampli-
tude at the node is zero, and at the antinode it is H.
,~,. H . 2nx . 2nt
'I'= sm-sm-
L T 15-1.2.4 Two waves of same period but different ampli-
or more generally: tudes traveling in opposite directions form a "partial
clapotis" and can be defined linearly by the sum of A sin
(x - Ct) + B sin (x + Ct). A partial clapotis can also be
A standing wave can be considered as the superposition considered as the superposition of a progressive wave
of two waves of the same amplitude and same period with a standing wave. A partial clapotis is encountered
traveling in opposite directions. In the case where the in front of an obstacle which causes partial reflection.
convective inertia terms are negligible, the standing wave
15-1.2.5 It will be seen that the wave velocity is in general
motion is obtained by a mere linear addition ofthe solutions a function of the water depth (see Section 16-3.2). Hence, if
for two progressive waves (see Fig. 15-3). The following the depth remains constant, wave crests remain parallel.
identity holds:
However, when the depth varies in a direction not parallel
H 2n H . 2n to the wave crest, different parts of the crest travel at
2 sin L (x - Ct) +2sm L (x + Ct) different speeds, and the direction of propagation of the
crest changes. This is the phenomenon of refraction which is
. 2n 2n encountered when a wave travels from one water depth to
= H sm L x cosT t
another water depth at an angle with the bottom contours.
Figure 15-3
Standing wave.
Part 3: Water Wave Theories
200
WAVE HE IGHT
WAVE CREST
Figure 15-4
BREAKWATER Wave diffraction.
mmzzmz{{;,-,-------~,~mmzzzz
15-1.2.6 The phenomenon of diffraction is encountered (2) when the pressure at the free surface given by Bernoulli
at the end of an obstacle (Fig. 15-4). It can be considered as equation is incompatible with the atmospheric pressure,
a process of transmission of energy in a direction parallel (3) when the particle acceleration at the crest tends to
to the wave crest although this definition oversimplifies a separate the particles from the bulk of the water surface, or
more complex phenomenon. The phenomenon of wave (4) when the free surface becomes vertical. Accordingly, the
diffraction is actually the result of phase differences be- following theoretical formulas are generally used:
tween the waves radiating from a line source, which is 1. In deep water (Michell limit): H/L < 0.142.
interrupted by the obstacle. It is an end effect. 2. In intermediate water depth (Miche formula) : H/L
< 0.14 tanh (2n d/L).
15-1.2.7 The breaking phenomenon is encountered at
3. In the case of a solitary wave, the maximum relative
sea under wind action (white caps), on beaches (surf), and in wave height: H/d < 0.78, where dis the water depth.
tidal estuaries (tidal bores) (see Fig. 15-5).1t is a shock wave
phenomenon which is also encountered in gas dynamics.
15-1.3 Translatory Waves
The breaking phenomenon is characterized by a high rat$!
of free turbulence and air entrainment associated. with a In a translatory wave, there is a transport of water in the
high rate of energy dissipation. Bores generated by wind direction of the wave travel. Some examples of such pheno-
waves breaking on beaches or by tides of high amplitude mena are tidal bores or moving hydraulic jumps; waves
in estuaries should be regarded as translatory waves generated by the breaking of a dam; surges on a dry bed;
rather than oscillatory waves. undulated moving hydraulic jumps; solitary waves; and
A number of equivalent definitions can be given for the flood waves in rivers. It is pointed out that oscillatory
breaking criterion: Breaking occurs (1) when the particle and translatory waves may sometimes look very much
velocity at the crest becomes larger than the wave velocity, alike and be treated mathematically by the same method.
Chapter 15: An Introduction to Water Waves
201
----
WHITE CAPS; LARGE WATER DEPTH.
Figure 15-5
Different kinds of wave breakers.
For example, a solitary wave, which is a translatory wave, is gentle slow return under a long flat trough between wave
characterized by a unique wave crest accompanied by a crests. A solitary wave motion always involves an important
sudden jump ahead by the water particles under the wave net mass transport. A cnoidal wave has a very small mass
crest. Cnoidal waves (discussed in far greater detail below), transport because of this return flow. From the mathe-
which are oscillatory waves, present very similar character- matical viewpoint, these two kinds of motion are of the
istics. However, in the case of a cnoidal wave, there is a same family, i.e., they are subjected to the same simplifying
Part 3: Water Wave Theories
202
_s_.w_.L----~ :1ft:-=-----
OSCILLATORY WAVE
Figure 15-6
Difference between an oscillatory wave
SWL and a translatory wave.
assumptions and they obey the same basic equations. The Although the relationships between the inertial terms and
solitary wave is a limited case of the cnoidal wave when the these three parameters are not simple, their r~lative values
wave period tends to infinity (see Fig. 15-6). are of considerable help in classifying the water wave
theories from a mathematical viewpoint.
15-2 Mathematical Classification For example, it is easily conceived that when the free sur-
face elevation decreases, the particle velocity decreases also.
15-2.1 The Significant Wave Parameters Consequently, when the wave height H tends to zero, the
15-2.1.1 In an Eulerian system of coordinates a surface convective inertia term, which is related to the square of the
wave problem generally involves three unknowns: the particle velocity, is an infinitesimal of higher order than the
free surface elevation (or total water depth), the pressure local inertia term, which is related linearly to the velocity.
(generally known at the free surface), and the particle In such a case, the convective inertia can be neglected and the
velocity. Since a general method of solution is impossible, a equations can be linearized.
number of simplifying assumptions have been made which Three characteristic ratios can be obtained from H, L,
apply to a succession of particular cases with varying and d. These are H/L, Hjd, and Ljd. The relative importance
accuracy. In general, the method of solution which is used of the convective inertia term increases as the value of these
depends upon nonlinear effects, i.e., the relative importance three ratios increases.
of the convective inertia terms with respect to the local In deep water (small Hjd, and small Ljd), the most
inertia. significant parameter is H/L which is called the wave
steepness. In shallow water the most significant parameter
15-2.1.2 However, instead of dealing with these inertial is Hjd which is called the relative height. In intermediate
terms directly, it is more convenient to relate them to more water depth, it will be seen that a significant parameter which
accessible parameters. Three characteristic parameters are also covers the three cases is (HjL)(Ljd) 3
used. They are:
ization; (2) power series; and (3) numerical methods. and 203
Statistical methods are also used to describe the complexity
of sea states or waves generated by wind action. _a
F = c -::-:--aF-----=:-:-
ot o(x - Ct)
small amplitude and small wave length in deep water. For As a result, the time derivatives can be eliminated and
the first reason they are called the "small amplitude wave replaced by space derivatives. Power series solutions are
theories." This is the infinitesimal wave approximation. also found in the case of periodic standing waves and irregu-
The linearized equations are so amenable to mathemati- lar waves.
cal solutions that the linear wave theories are used for an
extreme variety of water wave motions. For example, 15-2.2.3 It may happen that a steady-state profile does
some phenomena which are studied by this method are not exist as a solution, in which case a numerical solution
wave diffraction, waves generated by a moving ship, and where the differentials are replaced by finite differences, is
waves generated by explosions, even though they may often used. This occurs for large values of H/d and Ljd,
have large amplitudes. when the nonlinear terms such as pu(oujox) are relatively
l~rge by comparison with the local inertia such as p(oujot).
15-2.2.2 Solutions can also be found as a power series in This is the case of long waves in very shallow water.
terms of a parameter small compared to others. This small Of course, a numerical method of calculation can be used
quantity is H/L in deep water or H/d in shallow water. In for solving a linearized system of equations. For example,
the first case (development in terms of H/L, or Stokes the relaxation method is used for studying small wave
waves) the first term of the power series is a solution of the agitation in a basin. Also, an analytical solution of a non-
linearized equations. In the second case, the first term of the linear system of equations can be found in some particular
series is already a solution of nonlinear equations; these cases. Hence it must be borne in mind that these three
are the cnoidal waves. methods and the range of application which has been given
The calculation of the successive terms of the series is indicate more of a trend than a general rule.
rather cumbersome so that these methods are used in a
very small number of cases. The most typical case is the two- 15-2.2.4 Aside from the three previous methods which
dimensional progressive periodic wave. In this case, the aim at a fully deterministic solution of the water wave
solution is assumed to be a priori that of a steady-state problem, the description of sea state generally involves the
profile, i.e., a function such as F = f(x - Ct) where Cis a use of random functions. The mathematical operations
constant equal to the wave velocity. Cis also called the phase which are used in this case (such as harmonic analysis)
velocity. generally imply that the water waves obey linear laws,
The simplification introduced by such an assumption which are the necessary requirements for assuming that the
is due to the fact that principle of superposition is valid. Consequently, such an
approach loses its validity for describing the sea state in
oF oF very shallow water (large values of Hjd and L/d) and in the
ox o(x- Ct) surf zone.
Part 3: Water Wave Theories
205
LIMIT DEEP WATER WAVE
STOKES(4th
orde
~ =0.14
STOKES(3rd
order)
L()
0
0
STOKES(2nd
order)
DEEP
SHALLOW
N"
(.) WATER WAVE
-4-- -.wATER
Cll WAVE
-
en
......
Figure 15-7
Limits of validity for various wave theories.
L()
/UR= I ~
I -oi....J AIRY
THEORY
I I (LINEAR)
I
I
O.OOIO_L.0-:-1:-------L.!....__ _--l.._O.J...I_ _ _ _ _ _ _ _ _.J..__ ____.Il--_ _ _ _...JIO
rd2 (ft/sec2)
15-2.3.3 Figure 15-7 indicates approximately the range using various theories in various domains has not been
of validity of the various theories. This graph has been done so far; hence, such a graph is somewhat arbitrary
established for two-dimensional periodic waves such as and merely qualitative.
illustrated on Fig. 15-8, but it gives an indication for any
kind of water waves. Three corresponding values of the 15-2.4 The Two Great Families of Water
Ursell parameter have been shown. The graph is limited by Waves
a breaking criteria which implies that there is a maximum In hydrodynamics the water wave theories are generally
value for the wave steepness which is a function of the classified in two great families. They are the "small ampli-
relative depth (see Section 15-1.2.7). A comprehensive tude wave theories" and the "long wave theories." The
quantitative investigation of the error which is made by small amplitude wave theories embrace the linearized
Part 3: Water Wave Theories
(d) SOLITARY WAVE: LIMIT CURVE FOR CNOIDAL WAVE WHEN THE
PERIOD TENDS TO INFINITY.
theory for infinitesimal amplitude waves and the first 15-3 The Small Amplitude Wave
categories of power series, i.e., the power series in terms of Theory
H/L for finite amplitude waves.
The long wave theories embrace the numerical method 15-3.1 The Basic Assumption of the
of solution mostly used for the nonlinear long wave Small Amplitude Wave Theory
equations. It has been mentioned in the previous section that the
These two great families include a number of variations small amplitude wave theory is essentially a linear theory,
and some intermediate cases presenting some of the charac- i.e., the nonlinear convective inertia terms are considered
teristics of both families. For example, the cnoidal wave, the small. It is called the small amplitude wave theory because
solitary wave, and the monoclinal wave are considered as the equations are theoretically exact when the motion
being particular cases (steady-state profile) of the long tends to zero even if the convective inertia terms are taken
wave theories, because they are nonlinear shallow water into account. Indeed, in that case the nonlinear terms are
waves. infinitesimals of higher order than the linear terms.
It can be considered that there exists some arbitrariness This assumption is extremely convenient because the
in such classification. This arbitrariness is the heritage of free surface elevation can a priori be neglected; i.e., the
tradition, since the wave theories, as any theory, have been motion takes place within known boundaries. This assump-
developed in a haphazard manner. But it is most important tion is used in order to determine the zero wave motion
to understand the relative position of these theories with and such solution is assumed to be valid even if the wave
respect to each other, and their limits of validity. The small motion is different from zero.
amplitude wave theories and the long wave theories are Aside from this assumption, the motion is also most
now considered separately. often considered as irrotational. This assumption is
Chapter 15: An Introduction to Water Waves
208
Figure 15-10
Wave generated by a
free surface local disturbance.
general, these waves have a cylindrical symmetry but are theory. The theoretical wave patterns created by a moving
nonperiodic. One may consider that they have a "pseudo" ship are presented in Fig. 15-11. These are called Kelvin
wave period defined by the time which elapses between two waves.
wave crests. This period decreases with time at a given
location. Also, in general, this "pseudo" wave period 15-3.3 The Finite Amplitude Wave
tends to increase with the distance from the disturbance. Theories
Waves generated by disturbances often appear as a succes- 15-3.3.1 The solution for a progressive harmonic linear
sion of wave trains, the number of waves within each wave wave over a horizontal bottom is a sine function of(x - Ct),
train increasing as the distance from the disturbance so the free surface is perfectly defined by a sine curve. In
increases. shallow water, the crest has a tendency to become steeper
The average wave height also tends to decrease with and the trough flatter, as shown in Fig. 15-8c. In this case,
distance due to the double effects of the increase of wave the linear small amplitude wave theory is no longer valid.
length with distance and radial dispersion.
15-3.3.2 In the simple case of periodic waves, either
15-3.2.3 Finally, the wave motion created by a moving progressive or standing, the small amplitude wave theory
disturbance (ship or atmospheric disturbances) can also can be refined by taking into account the convective inertia
be analyzed by application of the small amplitude wave forces to some extent. It has been indicated in a previous
Figure 15-ll
Wave generated by a moving ship.
Chapter 15: An Introduction to Water Waves
section (Section 15-2.2.2) that this is done by assuming the It has been mentioned that in shallow water the impor- 209
solutjon for the motion is given by a power series in terms tant parameter becomes H/d instead of H/L for deep water.
of a quantity which is small compared to the other dimen- A power series in terms of H/d is most convenient and, in
sions. For example, in the simple case of a periodic progres- principle, would require fewer terms for a better accuracy.
sive or standing two-dimensional wave, it is assumed Such power series appear in the cnoidal and solitary wave
that the solution for the motion is given as a power series theories which will be discussed in the following para-
in terms of the wave height H (or of the wave steepness graphs (Section 15-4.2.1 ).
H/L, defined as the ratio of the wave height to the wave-
length L). For example, the potential function cf>(x,z,t) will 15-3.3.3 Once all the equations of motion and the
be written: boundary conditions at the free surface and at the bottom
have been specified, an infinite number of solutions may be
cf>(x,z,t)Hc/> 1 + H2 c/> 2 + H3c/> 3 + H4 c/> 4
=
found. These equations are not sufficient for determining
The first-order term Hc/> 1 is found from the linear small the wave motion.
amplitude theory by neglecting the nonlinear terms. The Two other conditions are required. One is on rotation-
other terms are correction terms due to the nonlinear ality and mass transport and is considered in Chapter 17.
convective inertia. These terms of the series are obtained It is also necessary to specify whether the wave motion
successively by recurrence formulas. A third-order wave should be a progressive wave, a standing wave, or a wave
theory, known as a third-order approximation, is a theory train. For example, in the first case, a solution for steady-
in which the calculation has been performed up to the state profile has to be found such that the solution appears
third power of the small quantity. In this case, a third-order as a function of (x - Ct) where C is the constant wave
approximation is velocity. In the second case, a mere addition of two periodic
gravity waves traveling in opposite directions can be used
Hc/>1 + H2 c/>z + H3c/>3
c/> =
in the linear case and for the first term of the power series.
In the case of a harmonic wave, c/> 2 and c/> 3 are sinusoidal However, higher-order terms must be found independently
functions of n(x - Ct) where n is an integer equal to the by recurrence formulas, established for the specific type of
order of the considered term. The 4>n are functions of the motion (progressive or standing).
relative depth d/L. In practice the complexity of the terms
c/> 2 , c/> 3, ... increases so much as the order of approximation
increases that calculation can rarely be performed at a 15-4 The Long Wave Theory
high order of approximation. The formulas for the high
order of approximation are so complicated that for their 15-4.1 The Basic Assumptions of the
application a set of tables obtained from a high-speed Long Wave Theory: The Long Wave
computer is required. Paradox
In engineering practice, the first-order wave theory is 15-4.1.1 The long wave theory applies when the relative
most often sufficient. However, higher-order wave theory depth is very small. Thus, the vertical acceleration can be
indicates some interesting trends for waves of large steep- neglected and the path curvature is small. Consequently,
ness (large H/L) in deep water. In very shallow water the the vertical component of the motion does not influence the
convective inertia terms are relatively high and the con- pressure distribution, which is assumed to be hydrostatic.
vergence of the series becomes very slow. The series are not However, contrary to the small amplitude wave theory, the
even necessarily uniformly convergent and the function of free surface is now unknown even during the first step of
relative depth d/L loses its meaning. the calculations. Also, the velocity distribution along a
Part 3 : Water Wave Theories
210 vertical is assumed to be uniform, or the particle velocity is may no longer be valid. Similarly, the breaking of a long
averaged over a vertical. (As in the case of the generalized wave on a beach will be predicted sooner than if it were
Bernoulli equation, a correction coefficient close to unity due to the change of bottom depth only.
should be included where quadratic terms appear. This Finally, it is realized that the long wave theory and the
refinement is neglected.) Because the equations are non- steady-state profile are two concepts theoretically in-
linear, the number of analytical solutions is limited to a few compatible, although steady-state profiles have been
particular cases. observed. This inherent deficiency in the long wave theory
While the small amplitude wave theory consists of is the long wave paradox and is also encountered in gas
finding potential function by analytical means, the long dynamics and nonlinear acoustics.
wave theory is most often treated by numerical methods or The two stabilizing factors which explain the existence of
graphical methods and by the use of a high-speed computer. steady-state profile are the vertical acceleration and the
bottom friction. They are now considered successively.
15-4.1.2 An error, inherent in the simplifying assump-
tions, is encountered systematically in the treatment of the
nonlinear long wave theory. The velocity of the "wave
15-4.2 Steady State Solutions
element" is an increasing function of the water depth such 15-4.2.1 If one takes vertical acceleration into account,
as (gh) 112 Therefore, according to the theory, the wave the pressure distribution is no longer hydrostatic. In
elements carrying the most energy should have a tendency to particular, due to the centrifugal force of water particles
catch up with the first wave elements ahead of the wave under a wave crest, the pressure at the bottom decreases
(see Fig. 15-12). A vertical wall of water should soon result, significantly (see Section 14-5.2).
forming a tidal bore. However, although this phenomenon Although nonnegligible, the vertical acceleration can
may actually occur physically, when it does occur, it still be linearized by assuming dwjdt ::::; owjot. Since the
happens much later than predicted by the long wave theory. vertical component of the motion is small, the convective
In particular, in the case of a wave which contains high terms w(owjox) and u(owjox) remain small. If this correction
space derivatives for oryjox and oujox, the long wave theory effect is taken into account in the long wave theory, the
motion becomes nonlinear horizontally and linear verti-
cally. Even if the nonlinear vertical components are taken
Figure 15-12 A physical illustration of the long wave into account in a solution obtained as a power series in
paradox. terms of H/d, they have to be introduced in high-order terms
I : 0 only for the sake of consistency in the approximations. If it is
assumed that the solution of the equation oflong waves with
a correction term for flow curvature is that of steady state,
this solution has to be defined as a function of (x - Ct).
Such solutions do exist. They are the solutions for solitary
waves and cnoidal waves (although in the latter case the
pressure is found to be hydrostatic at a first order of
approximation). In these cases, the value of the Ursell
parameter does not need to be small (see Section 15-2.3.1).
BORE In the case of very long waves where the vertical accelera-
tion and path of curvature are effectively negligible, the
choice of the significant length is quite arbitrary. This
Additional material from An Introduction to Hydrodynamics and Water Waves,
ISBN 978-3-642-85569-6, is available at http://extras.springer.com
Chapter 15: An Introduction to Water Waves
div V = 0 or -
ou + -OV + -OW = 0
ox oy oz
These assumptions result in a number of simplifications.
curl V = 0 ensures the existence of a single-valued
velocity potential function ljJ(x,y,z,t) from which the velocity
field can be derived. Thus, the potential function can
arbitrarily be defined as V = grad or V = -grad. The
latter definition is used in this chapter, i.e., u = - ol/Jjox,
v = - o/oy, w = - o/oz. The velocity potential function
y
z
Figure 16-1
Notation.
Chapter 16: Linear Small Amplitude Wave Theories
81
OZ z=-d-
-0
Figure 16-2 Notation.
16-1.4 The Free Surface Condition in the velocity at the free surface is now equal to the normal
Case of Very Slow Motion velocity of the surface itself. This gives with sufficient
In the case of slow motion the Bernoulli equation approximation
8
- - + - + gz =
p
f(t) ~~ =- ~~Lo
ot P
17 may now be easily eliminated from the dynamic and
becomes kinematic conditions. The derivative of 17 with respect to t
- 81
ot z=~
+ g1] = 0
in the dynamic condition gives
1]-! 8 I
- g ot z=~
01J/8t can be eliminated by equating the two above equa-
tions. This yields :
provided the function f(t) and any additive constant can
be included in the value of of/Jjot. [~~ + ~. ~:~1=0 = 0
Since the motion is assumed to be infinitely small, 17 which is called the Cauchy-Poisson condition at the free
may be written surface.
o1
-
OZ z=-d
= 0 and [a-Ot+ a]
2
2
OZ
g-
z=O
=0
Part 3: Water Wave Theories
( k2 - _E_) = 0 P"
g oz p
{ (~2 - )P
will give: 2
m2 = 0
o2U/ox 2 + o2Ujoy 2 d2 P/dz 2
U(x,y) P(z) (V 2 + m2 )U = 0
This may be written as The last equation is the well-known Helmholtz equation
(also called the wave equation) of mathematical physics.
V2 U P"
u p
16-2.2 Wave Motion along a Vertical
Notice that the functions of x and y are on one side of the
equal sign, while the functions of z are on the other. The The equation
variables have been separated.
It must be said that it was not certain at the beginning
that it would have been possible to separate the variables as
has been done. However, it will be shown later that this may easily be integrated, giving the general solution
process may be performed for solutions of V2 = 0. The
right-hand side of the above equation is a function of z.
The left-hand side is a function of x and y. Since x and y
where A and B are constants.
can vary independently of z and vice versa, the only way in
The boundary condition at the bottom,
which the function of x and y and the function of z can
always be equal (as stated by the above equation) is if the
left-hand side and the right-hand side are both equal to the 81
OZ
-0
z= - d -
same constant m 2 where m may be real or imaginary.
It will be easily seen that if m is imaginary there is no gives for any fixed value of x, y, and t:
physical meaning to the solutions in the case of wave
motion. Thus, m is chosen to be real and m2 is always dPI - o
positive. dz z= -d
Chapter 16: Linear Small Amplitude Wave Theories
When this boundary condition is applied to the solution where a, f3 are constant coefficients which depend upon 217
for P the result is the boundary conditions, k is the frequency 2n/T, and T is
mAe-md- mBe+md = 0 the wave period. When f3 = 0 and the coefficient rx is in-
cluded in the coefficient D, it is found
Hence
cfJ = D cosh m(d + z)V(x,y)eikt
Since there exist an infinite (but discrete) number of values
Consider the original solution, for kn and mn which satisfy the equation
p = Aemz + Be-mz k?; = mng tanh mn d
Multiply each term on the right by e-mde+md. Let a general solution for cjJ can be written as
Ae-md = Be+md = !D 00
218 Substituting the relationship k 2 = mg tanh md leads to This is the velocity potential function of a progressive wave
traveling in the OX direction.
,~,. _ _ !!:._cosh m(d + z) ( ) ikt If U = eimx, the velocity potential function of a wave
'f' - a . h d U x,y e
m sm m traveling in the opposite direction is obtained.
With the value of D that has been found, the expression for If the solution for U is:
P becomes
ak cosh m(d + z) ag cosh m(d + z)
P(z) =
m sinh md k cosh md or
Under these conditions the wave height at any point is 1 . .
2aU(x,y). U(x,y) is the relative value of the wave height with u= 2i (e'mx- e-mx) = sin mx
respect to a plane or a point where it is simply 2a.
then
16-3 Two-Dimensional Wave Motion ,~,. _ _ !!:._cosh m(d + z) {cos} k
'f' - a . h d . mx cos t
m sm m sm
16-3.1 Integration of the Wave Equation This is the velocity potential function of a standing wave.
The differential equation to be solved is (V 2 + m 2 )U = 0. If A' is different from B', a partial standing wave is obtained.
A general solution of this equation does not exist, but a In practice, the values for A' and B' are given by vertical
number of solutions may be found, corresponding to boundary conditions (wave reflection, etc.).
particular boundary conditions. In the case of a two- In the most general case of a two-dimensional irregular
dimensional wave such as motion encountered in a wave wave, as may be observed at sea, the velocity potential
flume. function is:
kn cosh mn(d + z) .
L
00
(..!:.._
ox 2 + m
2 )u = 0
When there are two waves only, traveling in the same
direction, the velocity potential function describing the
"beating" phenomena may be obtained easily.
Solving this, one finds that the solutions for U are given by
any linear combination of e-imx and eimx such as,
16-3.2 Physical Meaning: Wavelength
U= A'eimx + B'e-imx
It is easy to see the physical meaning of the coefficient
In particular if U = e-imx, then m. Since and consequently 11 is periodic with respect to
k cosh m(d + z) i(kt-mx) space, m = 2n/L and Lis the wavelength.
c/J= -a-
m sinh md
e The wavelength is given by
or
k2 = mg tanh md
= -a!!:._ cos~ m(d + z) cos (kt - mx)
m smh md and then
Chapter 16: Linear Small Amplitude Wave Theories
(-2n)
T
2 2n
L
2n
= - g tanh-d
L x = x0 - a
cosh m(d + z0 )
. h d
SID m
cos (kt - mx 0 )
219
220
10
20
X>
40
601
.....
..
u
0
Figure 16-3
0
Wavelength vs depth and period.
~ 6 ~~-#~~~~~~~~~---+----~----+-~~
0..
vr 2
L : ---
2Tr
vT 2 2Trd
L =- t a nh -
2Tr L
0 ~--~----~----~----L---~-----L----~--~
0 toO 200 300 400ft
10 ~ ~ ~ ~
WAVE
60LENGTH
ro ~ ~ 100 oo ~~ ...
The first-order potential function is wave height of the two progressive waves by measuring the
k cosh m(d + z) amplitude at the antinode (maximum) A and at the node B
<P= m sinh md (minimum). Then,
x [a 1 sin (kt + mx) + a 2 sin (kt - mx)]
The amplitude at the antinode is (a 1 + a 2 )
and at the
node is (a 1 - a 2 ). It is possible to determine the individual
Chapter 16: Linear Small Amplitude Wave Theories
60
221
55
~,,a-::;:.
50 /16
1---''4
~~
v'2-
v
45 ~ Vt'-
f--10
~ ~v-
---
f--9-
40 ~
f-B-
.,
u
~ 35 ~ ~~B;-
c,:j;j-
7
~~ ----
Cl
.,.:
~ __- ___.v -s::
/ Q
!::: a:
w
~----
u 0..
0 30
Figure 16-4 .J
w
Wave velocity vs depth and period.
>
w
_,......... ~:;':
5-
~ 25
~c' 1!...
~~~ / 2,.
~
20 4-
/
1~ L
~
f
15 3
~
10
5
r;
r;
2-
0
0 10 20 30 40 50 60 70 BO 90 100
DEPTH. ft
The reflection coefficient of the obstacle causing the partial 16-3.5 The Use of Complex Number
standing wave is: Notation
To show the great simplicity introduced by the operation
A-B W = <P + il/l, W is calculated to a first order of approxima-
R=--
A+B tion in the case of a two-dimensional monochromatic
progressive wave motion. First of all, it is convenient to
The envelopes of the crest and the trough are two sinusoids make a change in the origin of the vertical axis, and to take
of amplitude, (A - B)/2. the horizontal axis on the bottom, such as: z' = d + z.
Part 3: Water Wave Theories
222
...
~
X
1-
(!)
Figure 16-5
z
L1.l Wave length vs period and depth .
...J
L1.l
>
<(
:I: 16 15
100
o~--...JI0~--2~o~~~--~~~~s~o~~so~--~7o~~a~o--~9~0--~IO~O---I~I0---...~12-0~
0~~--~--~--+-~
WATER DEPTH, meters
With this new condition the velocity potential function The stream function l/1 is given by one of the following
could be written operations:
k cosh mz' ot/J ol/f or
ot/J ol/f
tjJ = -a- . cos (mx- kt) W=
ox
m smhmd
U=
ox oz' oz'
If This gives
A= -a-~~~
k1 l/1 = A sinh mz' sin (mx - kt)
m sinh md Now
then w= tjJ + il/l
tjJ = A cosh mz' cos (mx - kt). = A[cosh mz' cos (mx - kt) + i sinh mz' sin (mx - kt)]
Chapter 16: Linear Small Amplitude Wave Theories
I~ LI 2
'z
Figure 16-7
Standing wave.
Part 3: Water Wave Theories
224
Figure 16-8
Variation in amplitude in a
partial standing wave.
As an application of three-dimensional motion, the It is interesting to note that the equation of a two-di-
case of a standing wave in a rectangular tank is analyzed. mensional motion,
)u
The boundaries are shown in Fig. 16-9.
+ m2
-ocp = 0 or
oU O
- =
"
10r
{X = 0, =a
X
( : :2 = 0
o +-;:or+
(arz 1 a
2
)
mz u = 0
Chapter 16: Linear Small Amplitude Wave Theories
( : :2 + m2 )u = 0 (-az2 + --
8r
1a
r ar
+ m2) U = 0
e-imx =cos mx- i sin mx Progressive wave in the positive Converging wave (sink) (wave
OX direction (wave height height increases when r
=constant) decreases) when combined with
e+ikt
e+imx = cos mx + i sin mx Progressive wave in the negative Diverging wave (source) (wave
OX direction (wave height height decreases when r
= constant) increases) when combined with
e+ikt
eimx + e-imx Standing wave (horizontal H~1 > + H~> Standing circular wave (horizontal
cosmx=---- velocity = 0 when x = 0) 1 0 (mr)= 2
2 velocity = 0 when r = 0)
eimx- e-imx Standing wave (horizontal H~1>- H~z> Standing circular wave (horizontal
sinmx=---- Y0 (mr) =
2i velocity maximum for x = 0) 2i velocity infinite for r = 0)
Part 3: Water Wave Theories
226 given by the amplitude of U, is proportional to A and This formula is general and can be applied by inserting the
decreases as r 112 This may also be demonstrated by con- value of the potential function for any kind ofirrotational
servation of the transmitted wave energy (see Section 16-5). wave, linear or nonlinear. In the case of a linear periodic
progressive wave (see Section 16-3.1)
16-4.3 Wave Agitation in a Circular
Tank = -a}!_ cos~ m(d + z) cos (kt- mx)
The boundary condition for a circular tank is m smhmd
Inserting this value in the above expression leads to
a1 -0
ar r=R- _ p ~ Jt+T J~ 2 k 3cosh 2 m(d + z)
or Fav- a . 2
T 1 -d m smh md
x sin 2 (kt - mx) dz dt
where R is the radius of the tank. The solution tor U is After integration and the neglection of some high-order
U = Jn(mr) cos ne, where Jn is the Bessel function of order terms, this becomes
n. When n = 0, U = J 0 (mr) and the motion is a stationary 1 2 2md)
4pga C( 1 + sinh 2md
circular wave. Fav =
f+L
average energy flux per unit of wave crest through a fixed
vertical plane parallel to the wave crest is E= p f}!V 2 + gz) dz dx
1 ft+ Tfq (P TV2 + P + pgz )
Fav =T 1
-d U dz dt By inserting the value of V 2 = [(acpjax? + (acpjazf] and
neglecting some high-order terms, it is found that in the
i.e., by application of the Bernoulli equation where f(t) is case of a simple linear periodic progressive wave that the
assumed to be taken into account in a/at: potential energy EP equals the kinetic energy EK and the
F = -
1
p-
Jt+T Jq a a
- - dz dt
total energy per wavelength per unit length of crest is
av T t -d at ax E = EP + EK = !pga 2 L
Chapter 16: Linear Small Amplitude Wave Theories
Per unit horizontal area, this is Since m = 2n/L, dm = -2n dL/L 2 , and 227
dC
U = C- L dL
Dividing the energy flux by the stored energy gives the When the values C = (gT/2n) tanh (2nd/L) and L = CT
rate of propagation of energy: are inserted this becomes
U E = F av =
Eav
~ (1 +
2
2md )
sinh 2md U =
c(
2
2md )
1 + sinh 2md
i.e., U E = -!(gT/2n) in deep water and U E = (gd) 1 12 in It is interesting to note that this expression is the same
shallow water, which means that the rate of transmitted as the expression for the rate of energy propagation UE. This
energy is half of the wave velocity in deep water and equal stems from the fact that in the case of a wave train, there is
to the wave velocity in shallow water. no energy passing through the node of the wave train where
the wave amplitude is zero. So, the velocity of energy propa-
gation is that of a wave train or of a group of waves. For
16-5.3 Group Velocity this reason, the group velocity U is equal to the rate of
energy propagation. However, it is important to note that
Consider the linear superimposition of two progressive this statement holds true only in the case of a linear wave.
waves of the same amplitude and slightly different periods The values of U/C 0 and C/C 0 where C0 is the deep water
such as wave celerity (C 0 = gTj2n) is shown on Fig. 16-11. The
energy flux can now be expressed as a function of the group
Yf = a sin [mx - kt] + a sin [(m + bm)x - (k + 6k)t]
velocity U:
where bm and 6k are supposed to be small quantities. This
expression can be written as:
cos -!{6mx - 6kt) = cod bm(x - :~ t) from its value at B, and only the potential energy is trans-
mitted through the plane AB. Since the average value of
potential and kinetic energy are equal, there are only two
which travels at a speed U = 6kj6m. The wave velocity possibilities. Either one-half the wave energy is transmitted
C = k/m, and 6k = 6(mC), U = 6(mC)j6m. If these values with phase velocity or the total energy is transmitted at
are taken at the limit when 6m tends toward dm, U can be one-half the phase velocity, which is the deepwater group
written as U = d(mC)/dm. This is just U = C + m(dC/dm). velocity.
Part 3: Water Wave Theories
v--
1.0
228
1-"'
L_
v
__L
0.8 / Co
I
-------
u
/Co
f/
0.6
v
Figure I 6- I I
Relative wave velocity and group
(
0.4 velocity vs depth.
0.2
I
Thus, 229
1
4pga k tanh md[ 1 + sinh2md2md
2g J 1
= 4pgao
2g
k
16-6.2 Bottom Friction and Wave Damping
At a given location such as x = 0 the velocity at the bed is
and
ub
81
= ~ .
ka
h d cos kt
2a H 1 uX z= -d Sin m
2a 0 = H0 = {tanhmd[1 + (2md/sinh2md)]} 1i 2 The friction stress r for a turbulent boundary layer is
given by
= (~0 y 12 = K 5 r = pfubiubi
where .f is the friction factor (f ~ 0.015). The average work
where Ks is the shoaling coefficient. Inserting the expression done by this friction stress per unit area or as rate of energy
for LjL 0 = tanh (2nd/L) where L 0 is the deep water wave dissipation is
length L 0 = gT 2 j2n, it is then possible to calculate the
value of H / H 0 as a function of d/ L 0 only. The results of such
calculation are shown in Fig. 16-13.
\
1.3
\
1.2
\
1.1
0
--------
J:
......
J:
1.0
~r--
0.9
0.8
0 0.1 0.2 0.4 0.5
d/ Lo
Part 3: Water Wave Theories
230 Since H = 2a and k = 2n/T then, between wave orthogonals. The refraction coefficient K, is
(b 0 /b) 112 , where b0 is the distance between wave orthogonals
DJ = pf -T
1 fT (nH)3
. h3 d (cos kt) 2 [cos ktl dt in deep water. The combined effects of shoaling, damping,
3
0 T sm m and refraction are obtained by multiplying the shoaling
i.e., coefficient K., the damping coefficient K J and the refraction
4 z H3 coefficient K, so that
DJ = 3Pfn T3 sinh3 md
H = KsK 1 K,H 0
The variation of energy flux equals the rate of energy
Wave refraction diagrams giving distance between wave
dissipation, i.e.,
orthogonals are obtained by graphical methods or by
computer. Refraction effects remain small as long as the
water depth is larger than L/3. Therefore, wave refraction
or, H/H 0 = KsK 1 . The damping factor K 1 is calculations (and scale models) need to be done only over
bottom topography shallower than such a depth.
KJ = (1- ~ pg
Jd(x)DJ
-oo
dx) 16-6.3.2 The first method for drawing a wave refraction
diagram consists of calculating the wave crests first (Fig.
which is solved numerically over complex topography.
16-14). The wave crests are obtained by drawing the
envelope of circles from a preceding wave crest, the radius
16-6.3 Wave Refraction
of the circles being proportional to the local values of the
16-6.3.1 A wave orthogonal is the locus of points which wavelength, or one of its multiples.
define the minimum time of travel for wave propagation The second method consists of drawing the orthogonals
between two points. Wave orthogonals are usually normal directly by application of the Snell's law ofwave refraction:
to wave crests. They indicate approximately the direction
of propagation of the wave energy. When the wave arrives Cz sin IXz = cl sin IXt
at an angle with the bottom contours and refracts, the where the subscripts 1 and 2 correspond to two successive
distance between wave orthogonals b varies by a factor of values of the water depth.
b2 cos IXz
Not only is it possible to calculate the refraction co-
efficient from the relative distance of two wave orthogonals
bl COS IX1
but it is also possible to determine it from the curvature
IX is the angle of the bottom contours with the wave crest. of the orthogonals.
Subscripts 1 and 2 refer to the two referenced water depths
on both sides of the bottom contour. When the variation 16-6.3.3 In many cases refraction diagrams provide a
of wave height along a wave crest is small (which is not the reasonably accurate measure of the changes waves undergo
case of wave diffraction) it is generally assumed that the on approaching a coast. Quite often they provide the only
wave energy flux between wave orthogonals is constant, measure of these changes available. However, the accuracy
i.e., Favb = constant, and of data determined from refraction diagrams is limited
by the validity of the theory of their construction and the
tpgH 2 Ub = constant
accuracy of the depth data on which they are based. The
Hence, the wave height varies with a refraction coefficient orthogonal direction change is derived for the simplest
equal to the square root of the inverse ratio of the distance case of straight parallel contours and, although little error
Chapter 16: Linear Small Amplitude Wave Theories
231
SCALE
Figure 16-14 Wave refraction diagrams showing wave crests, wave orthogonals,
and relative wave heights along wave crests (shaded).
is introduced by bringing orthogonals over relatively limitation is inherent in the assumption that no energy
simple hydrography, it is difficult to carry an orthogonal travels laterally along a wave crest. No strict limits have
accurately on shore over complex bottom features. More- been set, but the accuracy of wave heights derived from
over, the equation is derived for small waves moving over orthogonals that bend sharply is questionable. A further
relatively flat slopes. Although no strict limits have been set, problem is the interpretation of crossed orthogonals. Such
accuracy as far as height changes are concerned cllnnot be cases would seem to lead to the formation of high, short-
expected where bottom slopes are steeper than 1/ 10. crested waves with breakers, and caustic, doubling wave
Model tests have indicated that direction changes occur crests due to nonlinear effects.
nearly as predicted even over a vertical discontinuity. In short, refraction coefficients which are quite different
Nevertheless, a small wave reflection takes place. A third from unity, say K, < 0.5 and K, > 2, must be doubted.
Part 3: Water Wave Theories
232 When they do occur as a result of one of the plotting pro- Inserting the expression for u (at a first order of approxima-
cedures mentioned, other effects such as diffraction may be tion), yields
occurring. Nonlinear effects also influence wave diffraction,
as can be seen by the doubling of wave crests at the ends of -p IT f~ (H- k cosh m(d + z) sm. (k t - mx))z dt dz
T 0 d 2 sinh md
breakwaters.
_ 1
- u;pgH
2[
1
4nd/L
+ sinh (4nd/L)
J
16-7 Wave Thrust or Radiation Stress Note that these two expressions are of the same order in
H 2 . Therefore, the wave thrust is
16-7.1 Definitions
The concept of specific force has been defined in Section 1
Sxx = u;pgH
2[
1 + 2 sinh4nd/L
(4nd/L)
J
14-2.1 as the sum of the momentum flux through a cross
section A and the pressure force acting on this section A which gives Sxx = ?6 pgH 2 in shallow water and Sxx =
per unit weight of fluid; i.e., / 6 pgH 2 in deep water.
vz p
[-+-A
g pg
J If we consider a wave traveling over a gently sloping bed
from deep water into shallow water to the point at which
the wave height becomes equal to the deep water wave
A similar definition is introduced m the field of water height, as seen in Section 16-6.1 the wave thrust has to
waves, the concept of" radiation stress." Radiation stress is, increase by a factor of 3. Therefore, the momentum balance
by definition, the average value of the sum (Pw + p v;) with requires an external force. This external force is obtained by
respect to time, integrated along a vertical plane of unit differences in the hydrostatic pressure. Hence, the mid water
width. The velocity component perpendicular to the con- level is set down below the still water level, as demonstrated
sidered plane is V,, Pw is the pressure fluctuation due to in the following section.
water waves around the hydrostatic pressure from the still
water level, i.e., Pw = p - pgz where p is the total pressure,
16-7.3 Wave Set-Up/Wave Set-Down;
and p v; is the momentum flux. When the plane is vertical Surf Zone Circulation
and parallel to the wave crest (V = u), the radiation stress is Let us consider the momentum balance in a slice of
water bounded by the free surface z = IJ, a gently sloping
bottom z = -d(x), and two vertical planes parallel to the
wave crest at x = x 0 , x = x 0 + dx. The average value of
which is still called the "wave thrust." ij = l/T J1J dt is defined and gives the variation of the mid-
water level with respect to the still water level. The total
16-7.2 Application to Water Waves forces exerted on the planes x = x 0 , x = z 0 + dx are the
sum of hydrostatic pressure and wave thrust. Thus,
If one inserts the expression of Pw obtained from the
linear wave theory, the corresponding integral has an
F xo = + [Sxx + !pg(d + ij) 2 ]
average value equal to zero. However the second-order d 1
term yields
Fxo+dx = - { Sxx + dx (SxJ dx + 2 pg
1 2 4nd/L
u;pgH sinh (4nd/L) x [(d + ij) 2 + ~ (d + ij) 2 dx ]}
Chapter 16: Linear Small Amplitude Wave Theories
Another external force is due to bottom pressure (since After breaking inception, the wave energy is dissipated; 233
the bottom is not horizontal). It has a horizontal component, therefore Sxx decreases and r; increases. This is the wave
set-up. The wave set-up is the elevation of the midwater
level which takes place in the surf zone. If one assumes that
all the wave energy is dissipated in the surf zone, the total
By application of the momentum theorem (Chapter 12), the force exerted on the beach (or on a wave absorber) is equal
momentum balance yields to the wave thrust in deep water, i.e.,
dSxx _)dry
Fxo+Fxo+dx+Fb~ dx +pg(d+ry dx~O
or, since r; ~ d, Therefore, it is seen that dissipating the wave energy does
dry 1 dSxx not nullify the average wave force. If the wave is reflected
dx ~- pgd dx instead of being dissipated, the wave force will be subjected
to large fluctuation but its average value, at a first order of
Since Sxx increases when the wave proceeds from deep water approximation, is zero.
to shallow water, it is seen that r; decreases, i.e., the mid water The concept of radiation stress also explains the occur-
level is set down (Fig. 16-15). rence of surf beats, i.e., the oscillations of water level on the
VERTICAL ELEVATION
(DISTORTED)
STILL
HORIZONTAL
DISTANCE
WAVE
THRUST
234 shore with a period equal to several times the wave period, 12,-------.-------.-------.-------.-----~
t VALUE
PROBLEMS
Figure 16-16 Hyperbolic functions us relative depth.
16.1 Determine the streamlines, the equipotential lines,
paths, and lines of equipressure (isobars) in a linear periodic
progressive wave.
16.4 Calculate the value of the wavelength Las a function of
the water depth din the case of a wave period T = 8 sec,
16.2 Determine the streamlines, the equipotential, lines,
10 sec, and 12 sec. Determine the minimum value of d/L
paths, and lines of equipressure in a linear periodic standing
when L can be considered as the deep-water wavelength,
wave.
and the maximum value of d/ L permitting the shallow
water approximation within 5% of accuracy.
16.3 Demonstrate byapplying the linear long wave theory
that the free surface of two-dimensional fundamental
16.5 Demonstrate that the velocity potential function for a
motion of a seiche in a basin in the form of a parabola
linear periodic progressive wave in deep water is
{d(x) = d0 [1 - (x 2 /a 2 )]} is a straight line. Demonstrate
that the free surface of the first harmonic motion is a k
= -a- e-kz cos (kt- mx)
parabola. m
Table 16-2 Periodic progressive wave, .first approximation: linear theory 235
Deep water Intermediate water depth Shallow ll"ater
d/L > 0.5 0.5 > d/L > 0.05 d/L < 0.05
C = [gL]''' = gT = f!.
gL 2nd]''' gT 2nd
Wave velocity C= [ -tanh- =-tanh- c= (gd) 1 12
2n 2n k 2n L 2n L
gT' gT 2 2nd
Wavelength L = - (Lft = 5.12 T 2 sec) L=-tanh- L = T(gd) 112
2n 2n L
Hg H g cosh m(d + z) Hg
cJ> = - 2 k exp (mz) cos(kt- mx) </> =- - -
2 k cosh md
cos(kt- mx) = - 2 k cos (kt - mx)
Circles of radius: R =!!. e+m" Ellipses of: H cosh m(d + Zo) Straight lines H T ( )'''
2 horizontal amplitude: - - 1!.
Semimajor axis: 2 --s1c-.nc-h-m--:d-
2 2n d
. . . H sinh m(d + z 0 )
Sem1mmor ax1s: - - - - - -
2 sinh md
H H H
Free surface '1 = 2 sin (kt - mx) '1 = 2 sin (kt - mx) '1 = 2 sin (kt - mx)
H H cosh m(d + z) H
Pressure p = -pgz + pg-
2 exp(mz)sin(kt- mx) (1)* p = -pgz + pg sin(kt- mx) P = - pgz + pg 2 sin (kt - mx)
2 cosh md
H sinhmz0
(2)* P = - pgz 0 - pg . sm (kt - mx 0 )
2 smh md cosh md
Energy per wave
length per unit of
length of crest
Hk H k cosh m(d + z) Hg
Potential function = - 2 ;;; e+m' cos mx sin kt = - --
2m sinh md
cos mx sin kt =---cos mx sin kt
2 k
Hg . H g cosh m(d + z) .
= - - - e+m cos mx Sin kt =--- cosmxs~nkt
2 k 2 k cosh md
X - Xo x- x 0 tan mx 0 H onzonta
. II"1nes _
H _
T (g/d) 112
Straight lines:--= -tan mx 0 Straight lines: - - = - -,----,.,--------,- sin mx
z- z 0 z - z0 tanh m(d + z0 ) of amplitude: 2 2n
H H H
Free surface 'f= --cosmxcoskt 'f= --cosmxcoskt 'f= --cosmxcoskt
2 2 2
Pressure p= -pgz-pg 2H e+""cosmxcoskt p= - pgz - pg 2H coshcoshm(dmd+ z) cos mx cos kt P= - pgz - pg lH cos mx cos k
ipg(~Y Lsin ipg(~Y Lsin
Energy per
wave length EK = 2 kt Ex = 2 kt EK = ipg(~Y Lsin 2 kt
and give the expression for the pressure p(x,z,t) and the free and demonstrate the expression:
surface. Demonstrate why the criterion d/L ::;; 0.5 is
generally considered as the limit of validity of the above gL 2rcd)
C = ( -tanh-
112
longest periods of free oscillation of a rectangular basin It will be assumed that this breaking criterion is still valid 237
10ft x 8 ft (3.048 m x 2.438 m) and 2ft deep (0.609 m). in the case of a wave breaking at an angle on a gently
sloping beach. Now, consider a periodic wave arriving
16.10 Determine the error on the value of C corresponding from deep water at an angle r:t. 0 with the bottom contours,
to the deep water approximation (C = gT/2n) when these bottom contours being straight and parallel and
d/L = 1/3. Determine the minimum value of d/L which defining a very gently sloping beach. Establish a method for
gives the same error for the shallow water approximation: calculating the angle of breaking wave crest r:t.b with the
C = (gd) 112 Determine the common value of d/L such that shoreline, the depth of breaking db and the wave height of
the error for the shallow water approximation equals the breaking H b as a function of r:t. 0 , the deep water wave height
error for the deep water approximation. H 0 , and the wave period T.
Finite Amplitude 17-1.1.1 This chapter deals with the problem of periodic
two-dimensional finite amplitude waves, in which non-
linear effects are partly taken into account by the use of
Waves power series. The subject matter has been introduced
briefly in Chapter 15 in the general introduction to water
waves. The purpose of this chapter is to present more
advanced considerations, which form the background
for this subject matter. However, the mathematical details
are beyond the scope of this book, so that only the main
results will be presented.
It has been shown that solutions can be expressed as
power series in terms of a quantity which is small compared
to the other dimensions. The small quantity used here is
H/L for small L/d; it is the most significant parameter
in deep water. H/d is used for large L/d; it is most significant
parameter in shallow water.
In the first case (development in terms of H/L), the first
term of the power series is obtained by application of the
linear theory. These are the Stokes (or "Stokesian") wave
theories. Many series are expressed in terms of 2a/L, where
a is a parameter such that
I(2a)n
n
1
-L
_H
--
L
(a = ~ at the first order)
while others are expressed in terms of H/L directly.
In the second case, the first term of the series is already
a solution of nonlinear equations. They are called cnoidal
wave theories because they are mathematically defined
by the so-called cnoidal functions. The definition of these
functions, as well as the development of these theories,
are beyond the scope of this book. These are the essential
basic principles.
at least five different irrotational Stokesian wave theories contains several theories which may not be identical.
at a fifth order of approximation; all of them are based Since all cnoidal representations are truncated series, the
on the same assumption, but the mathematical formulations order of approximation is important because high-order
are different. The free surface profiles are given for three of terms are generally significant.
them (8 = mx - kt): There are two types of cnoidal theories. The oldest
1J is intuitive in nature, while the newer theories are straight-
-
H
= a Lr
n
n cos n8 (17-1) forward and more rigorous. All are irrotational. The
1
primary intuitive theory is that of Korteweg and de Vries
m1J = rnA cos 8 + [B 22 + B 24(mAf](mA) 2 cos 28 (1895). The first and second terms of the series are deduced
+ [B33 + B 35 (mA) 2] (mA) 3 cos 38 but no scheme is presented for extension to higher-order
terms. The terms which are found are unique. More
+ B44(mA)4 cos 48 + B 55 (mA) 5 cos 58 (17-2)
rigorous theories have been developed. All are based on
where A is related to H/2 and the B coefficients are complex a perturbation expansion. Unfortunately, even though
functions of d/L. rigor prevails, the newer theories also diverge very rapidly.
It seems then that the high-order theory is not necessarily
!!__ = [1 + {3 13 (mA 0 ) 2 + {3 15 (mA 0 )4] cos 8 better than its lower-order counterpart.
Ao
+ [{322 + f324(mA 0 f]mA 0 cos 28
17-1.2 Vorticity and Mass Transport
+ [{333 + f33s(mA 0 f] (mA 0 ) 2 cos 38
+ /3 44(mA 0 ) 3 cos 48 + {3 55 (mA 0 ) 4 cos 58 (17-3) 17-1.2.1 The problem consists initially of solving a
problem satisfying continuity, momentum, and boundary
The {3 coefficients are complex functions of d/L. This conditions. The motion is assumed to be periodic and the
example of the mathematical representation for steady-state wave profile is assumed to be that of a steady state. However,
periodic waves over a constant depth illustrates the com- these assumptions are not sufficient for solving the nonlinear
plexity of a systematic comparison between wave theories problem. Two more conditions are necessary. This leads
supposedly developed with the same aim. The complexity to discussion of the problem of rotationality and mass
increases with the number of possible assumptions which transport (as they are related) and the arbitrariness of
can be used initially for developing a wave theory. water wave theories.
The arbitrariness in the calculation of wave motion
17-1.1.3 Consider further the case of irrotational waves. is inherent to the arbitrariness which prevails in the assump-
The values of the wave characteristics depend upon the tions which are used in the calculation of the mass trans-
number of terms chosen for the power series expansion, portation. The wave motion can be determined by assuming
either in terms of wave steepness H/L (Stokesian solution) that there is no mass transport at all. These are the closed
or in terms of relative height Hjd (cnoidal-type solution). orbit theories, such as the exact solution of Gerstner
The Stokesian power series solution is not uniformly (1809) in deep water and the power series solutions of
convergent, and the validity of the solution is lost when Boussinesq for shallow water. As a result of this assumption,
the relative depth djL tends to a small value (say, d/L < 0.1 the motion is found to be rotational and the vorticity is
for a fifth-order solution), since the coefficient functions in the opposite direction to the particle rotation, i.e., in
of d/L tend to infinity. opposite direction to what should be expected physically
The same occurs in the case of the cnoidal wave solution. under the influence of a shearing stress due to wind blowing
There is no unique cnoidal theory; rather, the literature in the wave direction (Fig. 17-1).
Chapter 17: Finite Amplitude Waves
~
wave steepness H/L to the second order of approximation,
one finds that the general relationship between mass
transport U and vorticity 2ry is given by
I
2 = H2 [au _m k sinh 2m(d + z)J
2
NET MASS TRANSPORT 11 4 oz sinh 2 md
where it is recalled that H is the wave height, m = 2n/L,
L is the wave length, k = 2n/T, T is the wave period, d
is the water depth, and z is the vertical coordinate.
In the case where U(z) = 0 (Boussinesq wave)
NO NET MASS TRANSPORT In the Stokesian wave theories (1'/ = 0) where one assumes
that the integrating constant is such that D = 0, one has
The wave motion can also be assumed to be irrotational, Figure 17-2 Mass transport as a function of vorticity.
in which case a mass transport distribution is found as a WAVE DIRECTION-
result of nonlinearity. These are the Stokesian wave theories
which include Stokes (1847), Levi-Civita (1925), Struik
(1926), and Nekrassov (1951).
Even though there is a given mass transport distribution
which is a function of the vertical coordinate, the integrating
constant is often determined by assuming that the average
mass transport D is nil for the sake of continuity, i.e., a
steady flow is superimposed such that
D= Jq U(z) dz = 0
-d
where U(z) is the mass transport.
Part 3: Water Wave Theories
242 which gives (assuming U= 0) Another condition is generally preferred. This condition
consists of assuming that the average momentum over a
U(z) = (Jl + 1)[U(z)l~=oJ
wavelength is zero. In this case, another expression for C
where J1 is an arbitrary coefficient. is found which results in a different mass transport. Thus,
The case where J1 = - 1 corresponds to the Boussinesq it is further realized that the calculation of wave theories
(closed-orbit) solution. J1 = 0 corresponds to the irrota- is subject to arbitrariness because of different assumptions
tional theory of Stokes. If J1 > 0, the average vorticity is that lead to different values of C.
in the same direction as the orbit direction, such as found
when a strong wind blows in the wave direction. The case 17-1.3 The Essential Characteristics of
where J1 < - 1 gives a negative vorticity and a negative Some Two-Dimensional Periodic
mass transport at the free surface which can be due to a Progressive Irrotational Wave Theories
wind blowing locally in opposite direction to the wave A brief review of some classical wave theories is now
propagation (a frequent nearshore occurrence). In the case presented.
where - 1 < J1 < 0, the vorticity is in the opposite direction
and the mass transport is smaller than in the irrotational 17-1.3.1 The linear theory of Airy in Eulerian coordinates
case. These facts are schematically represented in Fig. 17-2. gives the essential characteristics of the wave pattern in a
Due to viscous forces, the mass transport at the bottom simple formulation. The free surface is sinusoidal, particle
is always in the wave direction. Numerous experiments paths are elliptic and follow a closed orbit (zero mass
confirm this result, and the following value of the bottom transport), and lines of equipressure are also sinusoidal.
mass transport velocity can be derived: The terms in (H/L) 2 are neglected. However, nonlinear
effects such as the wave set-down can be determined from
s 2 mk
ub = u;H . h2 md
sm
this theory (Section 16-7).
The linear theory of Airy in Lagrangian coordinates
also gives elliptic particle paths, but the free surface and
17-1.2.3 Finally, it is recalled that the mathematical lines of equipressure are now trochoidal (as in the deep
representation of water wave motions are found as a water wave theory of Gertsner). (A trochoid is the locus
solution of a set of basic equations and assumptions. described by a point within a circle rolling on a straight line.)
These basic equations are the equations of continuity and The linear long wave theory is the same as the theory of
momentum, the solutions of which must satisfy given Airy where it is assumed that d/L is small. As a consequence,
boundary conditions. The first assumption is that of the formulas are simplified considerably. The pressure is
rotationality or mass transport as they are related. However, hydrostatic and the horizontal velocity distribution 1s
another condition is required. For example, for progressive uniform. The wave velocity is simply (gd) 1i 2 .
monochromatic waves, the solution sought represents
a steady state such that the potential function = 17-1.3.2 The theory of Stokes at a second order of
f(x - Ct), where Cis a constant equal to the wave velocity. approximation is characterized by the sum of two sinusoidal
In this case, the solution is unique. Although the steady- components of period T and T /2, respectively. As a result,
state solutions are of the same form, C is still undetermined, the wave crest becomes peaked and the troughs become
and for the determination of C, another condition is flatter. The wave profile can even be characterized by the
required. For example, it can be assumed that the average appearance of a hump in the middle of the wave trough.
horizontal velocity over a wave period at a given location Similarly, the elliptical particle path is deformed and tends
is zero, but the mass transport is then necessarily minimum. to hump under the crest and flatten under the trough.
Chapter 17: Finite Amplitude Waves
In this theory as in all the following wave theories, there necessarily a small parameter whereas H/L is always.) 243
is mass transport as a result of irrotationality and non- The series is nonuniformly convergent. The terms in
linearity. However, phase velocity, wave length and group (H/d) 3 are neglected. The results of this theory diverge
velocity are the same as in the linear theories. The terms significantly from experimental results.
in (H/L) 3 are neglected.
The theory of Stokes at a third order of approximation 17-1.3.4 The solitary wave theory of Boussinesq is the
is characterized by the sum of three sinusoidal terms of result of a purely intuitive approach. The vertical com-
period T, T /2, and T /3, respectively. The same logical ponent of velocity is initially assumed to be linearly distri-
results are found. Phase and group velocity exhibit non- buted from the bottom (equal to zero) to the free surface
linear corrections. The coefficients of H/L which are (equal to the linearized free surface velocity 81J/8t). The
functions of d/L tend to infinity when d/L tends to zero vertical distribution of horizontal velocity is assumed to
so that the theory cannot be used in very shallow water. be uniform.
(The series is nonuniformly convergent.) The terms in A correction due to path curvature (vertical acceleration)
(H/L) 4 are neglected. is added to the hydrostatic pressure. The equations of
The theory of Stokes at a fifth order of approximation motion are linearized vertically but remain nonlinear
is the sum of five sinusoidal terms. The coefficients of horizontally, i.e., convective inertia terms where the
(H/Lt are functions of d/L and tend to large values for vertical component of velocity appears are neglected,
n > 3 even sooner than in the case of the third-order but the product u(8uj8x) remains. The solution is then
theory, i.e., for larger values of djL. Consequently, the exact. As in any solitary wave theory, 1J has always a
fifth-order wave theory is less valid than the third-order positive value and there is mass transport equal to the
wave theory for small values of d/L and cannot be used when volume of the wave above the still water level. The terms
d/L < 0.1. The terms in (H/L) 6 are neglected. in (H/d) 2 are neglected. The wave velocity is [g(d + H)r1 2 .
The solitary wave theory of McCowan is more rigorous
17-1.3.3 The theory of Keulegan and Patterson belongs and satisfies the kinematic free surface boundary condition
to the cnoidal family of water wave theories. It follows the exactly. It corresponds to a higher-order solution than the
same physical approach as the theory of Korteweg and theory ofBoussinesq. The vertical distribution of horizontal
de Vries. From a purely mathematical view point, there velocity is nonuniform. The terms in (H/d) 3 are neglected.
are some inconsistencies as some third-order terms are 17-1.3.5 The stream function theory of Reid and Dean
included while some other second-order terms are ne- is also irrotational (V 2 l/J = 0) even though the use of the
glected; however, it gives relatively good results. The stream function allows the study of rotational motion.
horizontal velocity component varies with depth, while It is actually a numerical method requiring the use of a
the pressure is hydrostatic. computer. The solution is presented as a series of harmonic
The cnoidal wave theory of Laitone is mathematically functions with unknown coefficients determined numer-
rigorous. At a first order of approximation, the vertical ically in such a way that the deviation from the Bernoulli
distribution of horizontal velocity is uniform. There is no constant at the free surface is minimum.
mass transport. The terms in (H/d) 2 are neglected.
The theory ofLaitone at a second order of approximation 17-1.3.6 The formulation of God a is based on experimental
gives a nonuniform velocity distribution. There is mass results and assumes that the particle velocity at the crest
transport. The vertical distribution of mass transport of a limit wave is equal to the phase velocity. Accordingly
velocity is uniform. The second-order term becomes larger an empirical correction parameter is introduced in the
than the first-order term as Hjd increases. (H/d is not linear wave theory.
Part 3: Water Wave Theories
244 where (a,[J) are two parameters which define the mean
position of the considered particle. The particle describes
17-1.4 The Theory of Gertsner
circles of radius r = aemfl_ The wavelength is L = gT 2 j2n.
The rotational deep water wave theory of Gerstner is The free surface and the lines of equipressure are trochoids.
exact. It is a closed orbits theory with the rotation in The limit wave is a cycloid of steepness H/L = 0.31.
opposite direction to the particle path. Its interest is Figure 17-3 illustrates some essential characteristics given
mostly historical, since it dates to 1807, although it was by this theory. The free surface corresponds to f3 = 0 and is
also developed independently by Rankine at a later date. defined by the parametric equation
It gives a simple geometric representation of the wave
motion defined by the coordinates (x,z) of a given particle, x = a - a sin (kt - rna)
such as (see Fig. 17-3; note the z coordinate system is z=-acos(kt-rna)
positive downward):
The pressure on a particle whose mean position is f3 is
x = a - aemfl sin (kt - rna) z= f3 - aemfl cos (kt - rna) independent of time and is given by the expression
2n a2
k = 2n p = pg[J + ~ k 2 e2 mfl + constant
rn=y:
T 2
ORIGIN OF AXES
DATUM FOR
---------
(3
C Y = f3o- t'n X
~
I
DIRECTION OF WAVE /:; '(ROLLING CIRCLE
~ DIRECTION
-
OF
II:" ' RADIUS y = f3 0
ROTATION OF
GENERATING CIRCLE
z
Chapter 17: Finite Amplitude Waves
The pressure is constant if {3 is constant, and thus all particles in Figs. 17-4 and 17-5. A comparison of various free surface 245
at the same mean level have the same pressure as they move profiles given by theories and experiments is also presented
about; this pressure is the same as the pressure on that in Fig. 17-6. All these comparisons are done for relatively
particle in its still position. The vorticity is shallow water waves of large relative amplitude. It is
2m 2ka 2e 2mfJ
seen that the linear theory is always good at the bottom
2ry = 1 - m2e2mfJ
where the boundary condition is linear and the convective
effects are relatively small. The linear theory loses its
Finally, the energy in a wavelength is validity toward the free surface. However, in general, the
nonlinear theories are not better than the linear theory.
Even though the numerical method of Reid and Dean
is not shown in this figure, it has been proven to be as good
as or better than the other theories when compared to
laboratory experiments. The empirical formulation of
17-1.5 Determination of the Domain of
Goda is still the best. Due to its simplicity, it appears that
Validity of Finite Amplitude Wave
an empirical relationship of this kind is the best method to
Theories
recommend from an engineering standpoint. This leads
17-1.5.1 Knowledge of the features ofthese wave theories us to discuss the selection of the "best" theory for practical
is essential to the engineer interested, for example, in purposes.
calculating the forces on submerged structures such as
17-1.5.2 In a narrow two-dimensional wave tank, viscous
pilings. For this purpose, he needs reliable expressions
effects in the boundary layer near the bottom induce
for the velocity field within the wave. Upon choosing a
vorticity which diffuses upward. Also, the fact that the
suitable design wave (that is, mean water depth, wave
wave tank is limited causes a return flow which nullifies
height, and wave period), he then proceeds to select a
the total mass transport. (If the wave tank is large, transverse
theory to describe that wave. This choice is not easily
effects take place, and the return flow meander.) The
made, since he must evaluate the applicability of at least
theoretical approach which takes into account viscous
a dozen theories. It may be hoped, for example, that a
effects gives a mass transport which is relatively well
theory developed to a fifth order of approximation is more
verified by laboratory experiments. Figure 17-7 shows
accurate than its lower-order counterparts, which may or
qualitatively the results of such theory.
may not actually be the case. Unfortunately, the range of
validities of the various wave theories is not well defined. 17-1.5.3 Under storm condition, the wind stress induces
(An attempt to classify them has been presented in Fig. 15-7.) a large mass transport and vorticity in the wave motion.
Also, only a few experimental results are available. Some This mass transport remains in a thin upper layer at the
are presented here. beginning of the storm and slowly diffuses downward
In this comparison with experiments, one can choose the under the influence of the turbulent fluctuations (mostly
horizontal particle velocity under the crest as an important due to white caps). At the same time, the mass transport
feature, since, in applications, velocity is generally most near the surface, which is in the main direction of the wave
critical and the velocity under the crest is the greatest travel at the beginning of the storm, tends to be subjected
attained at any depth. Furthermore, a theory which to the Coriolis effect. Therefore the mass transport near
prescribes the velocity field well is constrained to be good the free surface is at a small angle with the wave direction
for other features, such as accelerations and pressures, and has a direction which changes with depth. Eventually
a priori. Velocity profiles under the wave crest are shown the mass transport results in a storm surge which, in
Part 3: Water Wave Theories
LONG WAVE
1st CNOIDAL
2nd CNOIDAL
,
' v~'- 2od STOKES
,+
N
~
/I
3rd STOKE' I T = 2.2 sec
H = 0.323 ft; 10.60cm
I I D = 0.619 ft; 20.31 em
I I
I I
0~----~----~~~~--~-UL---~~--~~L----L----~----~------L---~
0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6 4.0 4.4
u, ft /sec
0 30 60 90 120
em/sec
addition to being influenced by Coriolis acceleration, is theory. The nonlinear corrections are meaningless for an
also subjected to coastal boundary effects. The storm engineer confronted with decision on what his structure
surge calculation method is based on the nonlinear long should look like. Failure of Texas towers may be attributed
wave equations (see Section 18-1.2.4) independently of water to the neglect of wind-induced mass transport in water
wave effects. Therefore, it is seen that storm surge calculation waves.
and nonlinear wave theories are two separate approaches The correct theory is a rotational wave theory with mass
for what is actually a unique phenomenum resulting from transport function of wind forces, duration of wind, and
air-sea interaction in shallow water. Actually, it is a fallacy fetch (length of water surface over which the wind exerts
to use irrotational or closed-orbit nonlinear wave theory its action), as done in wave hindcast methods. The descrip-
for calculating wave forces on piles, for example. The tion of such a method is beyond the scope of this book.
wind-induced mass transport near the free surface could However, a possible approach consists of finding a stream
be as high as 80% of the particle velocity given by the linear function solution of V2 t/l = f(t/1).
Chapter 17: Finite Amplitude Waves
1.4
y 247
/
1.2
/
/
/
4- 2nd CNOIDAL
1.0
/
McCOWAN!/
"0
......
~
0.8
/1
II
N
+
"0 1
y :
0.6
T = 1.16 sec
0.4 /1 I H =0.255ft(7.7cm)
0 =0.587ft(17.89cm)
I 1st CNOIDAL
I I
0.2
BOUSSINESQ I
I
u , ft/sec
0 30 60 90 120
em/sec
17-1.6 Limit Wave Steepness wave. Rotationality at the crest in the direction of the
wave travel such as that due to a generating wind will
Although it is beyond the purpose of this book to deeply reduce the limit wave steepness to a smaller value (see
analyze limit wave steepness and the phenomenon of Fig. 17-8). A deep sea wave steepness larger than 0.10 is
wave breaking, the relationship between the rate of rota- rarely encountered.
tionality and the limit wave steepness is worthwhile Rotationality in the opposite direction will theoretically
mentioning. increase the limit wave steepness. Such a case can be
It has been mentioned (Section 15-1.2.7) that wave- observed near the coasts when the wave travels in the
breaking inception will occur when the wave profile opposite direction to a wind blowing offshore. At the limit,
reaches a limit wave steepness H / L Imax. This limit steepness according to the closed-orbit Gertsner theory, the maximum
is theoretically 0.142 for a deep water irrotational periodic limit steepness is 0.31, but the rotationality at the crest
Part 3: Water Wave Theories
03.---------------.--------------,,--------------,--------------~--,
10
,:::0.1
~
E
(,) z
Q
1-
Figure 17-6
0 ~
...J 0
Free surface elevation
, ... __ . . . . --
w
" -------
over one-half wavelenyth.
~~
T = 1. 6 sec
\.--'' -------
H = 0 .27 1 It ( 8 . 26 em) ... ~7 -
~ . =---
0 = 0. 586 It ( 17.07 em)
----._
~-
-s
2 3 4
DISTANCE FROM CREST , x , ft
0 30 60 90 120
em
is then infinity and in the opposite direction to the wave already been subjected to investigation to some extent.
travel. It is evident that this result of the Gertsner theory However, a general theory of irregular waves with an
has no physical significance. arbitrary rotationality or mass transportation and quadratic
It is seen how important it would be to establish a bottom friction, valid for any wave height, any wave period,
general rotational wave theory and to relate the rota- and any water depth, would be preferred. Then it will
tionality and mass transport to the wind action and bottom remain to calibrate some parameters as functions of wind
friction. The effect of viscous friction at the bottom has stress, wind duration, and fetch length.
Figure 17-7
Periodic progressive wave with
mass transportation effect.
LARGE + SMALL ~
Chapter 17: Finite Amplitude Waves
WAVE TRAVEL
WAVE TRAVEL 249
WIND
NO WIND
17-2 Irrotational Waves: Methods of The free surface conditions are expressed in terms of
Calculation and Results their values at the still water level (z = 0). Developing
(x,y,O + IJ,t) in terms of power series in 11 gives
17-2.1 General Process of Calculation
It is assumed that the two unknowns, the velocity
,~,. A.(
'~'(x,y,IJ,t) = '~' x,y,O,t) + 11 [81/J(x,y,O,t)]
az +
potential and the free surface elevation IJ, may be trans-
formed into the following power series expressions with or
respect to a parameter a. This parameter is chosen arbi-
trarily but has the dimension of a length, usually taken to [Jz=~ = [Jz=o + 11 -8[8] 2 18 ]
+ -1J2 , 1-
2
8 2 +
Z z=O L Z z=O
be a half-wave height.
When this expression for is introduced into the free
= a1 + a2z + a 33 + + ann + surface conditions given in Section 16-1.3.2, one obtains
IJ = GIJ1 + a21J2 + a31J3 + ... + an!Jn + ... the kinematic condition
When this expression is substituted in the continuity
relationship V2 = 0, it is seen that each of these terms
n is an independent solution of the Laplace equation
V 2n = 0. Each 1/Jn also satisfies the fixed boundary con-
dition 8nf8n = 0 and the free surface condition.
Part 3: Water Wave Theories
+ { ~ ( + 11 ~~ + .. )}
2
] + gl1 = 0
Now the problem is completely formulated. When the
expressions for and 11 are substituted in the above free
surface conditions, calculations can be completed.
a
ax =
'
P(z)F (x - Ct)
u-
( -C- C) 2
+ C2 +
w2
cz
2g1]
= constant
Hence it is seen that The constant can be taken equal to unity. It is seen that the
motion could be considered as a steady motion in a new
a = - Ca- =+Cu
- system of relative coordinates. In this system of coordinates,
at ax the origin of the OX axis moves at the wave celerity C
Introducing these relationships, it is seen that the Bernoulli (the Rayleigh principle). The free surface and the bottom
equation in the case of progressive waves takes the form are streamlines in this moving system of coordinates.
gL 2nd] 1 12
Wave velocity
2rr
C = [ -tanh-
L
Third order - [ 1 + (H
c-_ {gL 2rr)']}'! 2
---
2n 2 L
Midlevel elevation Ll
A = 1112- (-H2)2
~ = m(q)
2
t;n~ md (l + 2 sin~2 md)
Mass transport (H)'
(nonclosed orbits) m2 (1) No return flow: U = m2 (ii_) 2 C cosh 'm(d + z)
U = 2 Ce+ 2 m'
2 2 sinh 2 md
(2) Average mass
transport nil: U = m 2 (H)'
- C~----
1 ( cosh 2m(d + z)- -
sinh 2md)
-
2 2 sinh 2 md 2md
Part 3: Water Wave Theories
Figure 17-9
Flow through porous medium
h(x, t)---> still water level (SWL).
t INCREASING Water waves: IJ(X, t) oscillates.
DATUM
Chapter 17: Finite Amplitude Waves
- -o + -1
ot 2
[(o) (o)
-
ox
2
+ -
oz
2
] + grl = f(t)
z n
(o) + (o)
media.
At the free surface z = h(x,t) = [x,h(x,t),t]. Differ- o =
_!__
2 2 o
entiating h with respect to t and x successively (* means K ot ox oz oz
Part 3: Water Wave Theories
254 which can be written as Table 17-2 summarizes the main equations for the
!__ o =
K ot ox
i_ ( o) + ~ ( o) _ o
ox oz oz oz
formulation of a two-dimensional wave and flow through
porous medium.
Table 17-2 Equations for two-dimensional waves and flow through porous mediums
fu
2
] _ a
fu
or
~: ah a [ oh]
K. at= ax h ax
(Dupuit approximation)
Chapter 17: Finite Amplitude Waves
PROBLEMS
due to ground vertical displacement. It is then sufficient to be unidimensional. (In the long wave theory, the vertical 257
to replace 01]/0t by 8(17 + d)jot in the above expression to component is not considered as one dimensional.) Therefore,
obtain the continuity equation. (The momentum equation-
neglecting friction- remains the same.) p(au + u au + w au) = - op*
Ot ox oz Ox
18-1.1.3 In a river or an estuary with gentle variation of
Adding the zero quantity u[(oujox) + (owjoz)], replacing
width l(x) the continuity equation is (A = hi)
p* by its hydrostatic expression as seen in the previous
section, and integrating gives
and the storm surge is then considered as a succession of they are also considered valid for unsteady motion, such
quasi steady motion. Adding also the Corio lis components, as those due to flood waves or tidal motion in an estuary.
friction forces and body forces due to the sun-moon However, one must realize that the practical use of
attraction, the equations of tidal motion are obtained. such empirical formulation is only due to the lack of
In the study of oceanic tide, the ground displacement due further information on the friction factor for unsteady
to earth tide (earth deformation) may also need to be taken motion. The application of the Chezy coefficient to unsteady
into account. motion is valid provided the velocity distribution is
influenced by the friction up to the free surface as in the
18-1.2.5 For simplicity, consider the one-dimensional case of a steady flow. This occurs when the expression for
motion. Since the wave motion at the bottom is far from the boundary layer thickness, which increases with the
being negligible, the shearing force due to friction has in wave period, becomes much larger than the actual depth h.
practice a relatively great importance in long wave theory. Still, density effects in a marine estuary have a nonnegligible
The long wave equation should be written (Fig. 18-1): influence on the distribution of friction forces. They can
only be known by analyzing field records.
au au aYf r
-+u-=-g-----
at ax ax p(d + Y/) 18-1.2.6 When the zero quantity g[d(d)jdx] gS where
In general, r will be assumed to be quadratic, i.e., r = pfu 2 . S is the bottom slope (which can be positive or negative
The coefficient f is assumed to be independent of the depending upon the motion direction with respect to the
variation of u with respect to time. It is assumed that X-axis) is added to the long wave equation, (d + 17) is
f is the same as if the motion were steady. Consequently, f replaced by the total depth h, and u is replaced by V, the
can be expressed in terms of the Chezy coefficient Ch: following results:
f = g/C~. The long wave equation now becomes (see av av ah g VIVI
Section 14-1.1): -+ V-+g-=+gS----
at ax ax - c~ h
au au a11 g u Iu I
at + u ax = -g ax - c~ (d + Yf) Such an equation is known as the equation of Barre de
St. Venant, which is used in the study of flood waves or
It has been seen that the friction coefficients such as f, tidal estuary in connection with the continuity equation
Ch, or n are used for gradually varied flow. Most often, given in Section 18-1.1.3.
Figure 18-1
Notation for long wave theory.
Chapter 18: The Long Wave Theory
Consider the points 1, 2, 3, 4 in a T-X diagram separated by 18-3.1.2 The error and limit of validity of numerical
intervals L1x and L1t, respectively (see Fig. 18-2). Knowing procedures is now discussed. The transformation of a
the value of '1 at points 1, 2, 3, the value of u at point 4 differential equation into finite difference involves a
is obtained from the equation: systematic error. Indeed, it is known that by developing
a differential term into finite difference by a Taylor expan-
L1u L11J sion yields
-=-g-
M L1x
oF L1F L1x , L1x"- 1
The known values are used to find u4 : ox = L1x - 2 f (x) ... - -n-!- J<n>(x) ...
T T
Figure 18-2
-j~x 1-
X X
Two kinds of
4 X X
finite difference schemes.
)( X X l(
2
~-------------------- X ~-----------------------4--x
SQUARE MESH STAGGERED MESH
Part 3: Water Wave Theories
262 In brief, the numerical implementation of the long wave and then differentiating so that oc 2 jo* = co(2c)jo*, and
theory always involves an inherent error, which increases dividing the second equation by c yields
with time andjor distance.
au au a2c
The study of the propagation of a bore, or an undulated at + u ax + c -a; = - gS
wave, or the wave created by the breaking of a dam over a
long distance is unreliable even if the stability criteria is a2c au o2c
satisfied because of the cumulative effect of the error. -+c-+u-=0
at ax ax
However, the study of a tidal wave in an estuary, or even
of a flood wave with gentle variation of depth, is possible. Adding and subtracting these two equations gives
Similarly, the propagation of a breaking wave (bore) over a
ot (u 2c)
a
+ (u c) ox (u 2c) =
a steep beach, i.e., over a short distance because of the -gS
steep slope, may give reliable results.
It is now recalled that the total derivative of an expression
18-3.2 The Method of Characteristics A(x,t) with respect to time is
18-3.2.1 The characteristic equations are now established. dA(x,t)
oA dx oA
Consider the long wave equations in the following form: ~=at+ OX dt
au au ary Thus, the left-hand side of the above equation is the
Momentum: - + u- = -g-
at ax ax total derivative
u = 0 is dxjdt = 0 + c, i.e.,
(0 + 2c)l 1 =o = (0 + 2c)lt=t
and
~I c = dx = (gd)1 f2
dt
'--------x '--------.-x
18-3.2.3 In general, the variation of u and 1J (or c) are
Figure 18-3 Advancing and receding gentle enough in order that a finite interval method can be
characteristics. applied. The basic characteristic equations are then written
for finite intervals .1x, 11t:
(d/dt)(u 2c) = Oi.e.,u 2c = constantvaluesK 1 andK 2
.1x
along lines of slope dx/dt = u c. Consider the case .1(u 2c) = -gS .1t along ,1( =uc
of a vertical wall of water which is suddenly released as in
the case of a dam breaking (see Fig. 18-4). It is seen that when The time history of the wave evolution can then be
t = 0, u = 0, and c = (gd) 1 ' 2 , where dis the water depth and determined step by step as follows. The values of u(x,t 1 )
the constants K 1 and K 2 are and 1J(X,t 1 ) are given for a wave at a timet = t 1 (see Fig. 18-5).
u + 2c = K 1 = 2(gd) 112 The values of u and c are calculated at regular intervals
u- 2c = K 2 = -2(gd) 112 .1x. The characteristic line of slope .1xj 11t = u 1 + c 1 is
drawn from point 1 and the characteristic line of slope
At time t = t 0 , the water wall collapses. At the down- .1x/M = u 2 - c 2 is drawn from point 2.
stream water tip: 1J = - d and c = 0. Hence the speed of the Their intersection at point 3 defines x 3 , t 3 graphically.
water tip is dx jdt = u( + 0) and Then, by applying the characteristic equation along these
(u + 2c)lr=o = (u + 2c)lr=r lines, the values ofu 3 and c 3 (and consequently 1} 3 ) are found
from the two equations
becomes 0 + 2(gd) 1 ' 2 = u.
The water tip travels at a speed 2(gd) 1i 2 . The speed of the u3 + 2c 3 = u1 + 2c 1 - gS 1(t 3 - t 1 )
rarefaction wave traveling upstream where 1J = 0 and u3 - 2c 3 = u2 - 2c 2 - gS 2 (b - t2)
Figure 18-4
Notation for the dam break problem
AB = 2CD.
Part 3: Water Wave Theories
Figure 18-6
Domain of influence
and domain of dependence.
Chapter 18: The Long Wave Theory
M
'74 = IJz + 2 Llx [u(d + '7)1 3 - u(d + '7)1 1]
and from the momentum equation one has:
Part 3: Water Wave Theories
266
6 7 8 9 10
10
, ,woz~
Figure 18-9 Application of the method of characteristics to
a wave breaking on a slope.
The calculation will proceed by calculating u 5 and IJs The variation of u and c with respect to x and t also has
from points 2, 3, 6, and so on. to be gentle enough in order that the finite difference terms
have a value close to the differential.
18-3.5 The Stability Criterion
The error made in transforming the differential equation 18-4 On Some Exact Steady-State
into a finite difference equation may cause a cumulative
Solutions
error which may blow up as t = L M increases. The finite
difference method is stable provided the point (x,t) 4nder In general, an unsteady-state solution is a typical
consideration is within the domain of dependence formed characteristic of the long wave theories, i.e., the wave
by the intersection of the characteristics lines from the profile changes its shape as the wave proceeds. This fact
points which are used for its determination. is inherent in the long wave equations. However, the
Consequently, since in general (u + c) 1 > (u - ch (see insertion of bottom friction and a vertical acceleration term,
Fig. 18-11), the criterion for stability is /).t < (f).xju +c). even approximated, into the long wave equations permits
Chapter 18: The Long Wave Theory
l~
10
II II I I L! I I I v
I
6
I
LUJ 11-~ 15 T=
4
.
5 I I
J_LLJ},;. 11/<~ I
~ I I' SPILLING BREAKER ILINI
~~ i
10 -9 -8
~I
-7 -6 5 -4 -3 -2 0
I
4 5 6 7 9
X
the finding of some special steady state solutions. They are Figure 18-11 Notation for application of the
the "monoclinal" wave, the solitary wave, and the cnoidal square mesh method.
wave. Only the first two cases are analyzed in this book.
It is recalled that the assumption of a steady state solution
results in the search of a function such as
:} = f(x- Ct)
where C is a constant. Consequently, 6
a
-=-C-
a
ot ox
Part 3: Water Wave Theories
l
oh + o(hu) = 0
at ax 18-4.2 Solitary Wave Theory
ou + u ou + g oh = gS _ jf_ u Iu I
Consider the long wave equation for a horizontal bottom
ot ax ox c~ h with a vertical acceleration term as it has been established
in Section 18-1.3:
Inserting the relationship o/8t = - C(ojox) in the continuity
equation gives (u - C)(8h/8x) + h(oujox) = 0 which inte- ou au 01] d + 1] 03 1]
grates to (u - C)h = A(constant). The momentum equation at + u ax + g ox+ -3- ot 2 ox = 0
becomes
01] + o[u(d + 1'/)] = 0
(u _ C) au
ox
+ goh =
ox
g(s _ ~)
C~h
at ax
Figure 18-12
M onoclinal waves.
Chapter 18: The Long Wave Theory
When the free surface curvature is negligible, the value for C After separation of the variables, it can be integrated in the
becomes form
18.3 Consider a square mesh of mesh size A in a X, Y tion of two-dimensional long waves into a finite difference
system of coordinates. Demonstrate that the linear long system corresponding to intervals Ax, At forming a square
wave equation for periodic motion for a finite difference mesh. Do the same in the case of a staggered mesh.
scheme follows the relationship (m = 2n/L):
18.9 Demonstrate that the equation of Barre de St. Venant
-4</Ji,j + </Ji,j+ 1 + <Pi+ 1,j + </J;,j-1 can be written:
(Am) 2
+A. gd 'f'l,J-=0
1 +--A..
'f/1- ,]
18.4 Let us consider the linear long wave motion defined by where
the free surface equation, such as: m = 2n/L, k = 2n/T
a= [ Q2 _ g~ 3 ]
ry(x,y,t) = cos kt cos mx + sin kt cos my
This defines a motion obtained by the superposition of f3 = QA
two standing waves at right angles. y = A3
1. Establish the equation of the lines of constant wave
amplitude. F = A3 [~!!!_- S- IQJQIJ
g [2 dx C~A
3
2. Determine the points of maximum amplitude.
3. Is there any amphidromic point (point of zero Give the definition for .
amplitude)?
Describe the motion around it.
4. Draw a sketch of the curve of constant amplitude REFERENCES FOR PART THREE
within a square defined by 0 < x/y < L/2.
Basset, A. B., A Treatise on Hydrodynamics, Vols. I and II. Re-
printed by Dover Publications, Inc., New York, 1961.
18.5 Establish the equation of motion for periodic linear
long wave in a corner defined by a depth d = est, and width Beach Erosion Board, A Summary of the Theory of Oscillatory
Waves. Technical Report No. 2, U.S. Government Printing
b = b0 x. Office, 1942.
Biesel, F., Equations generales au second ordre de Ia houle irreguliere.
18.6 Consider the energy diagram as shown on Fig. 14-5. La Houille Blanche, May 1952.
Indicate the modification which needs to be made to such Birkhoff, G., Hydrodynamics. Princeton University Press, 1950.
a figure in the case of unsteady motion. Bouasse, H., Houles, rides, seiches et man!es. Librairie Delagrave,
Paris, 1924, pp. 92-145.
18.7 Demonstrate that the long wave equations can be Boussinesq, J., Theorie de !'intumescence liquide appelee onde
transformed as u + 2c = constant and u - 2c = constant solitaire ou de translation se propagant dans un canal rectangu-
along a line of slope laire. Institut de France, Academie des Sciences, Comptes
Rendus, June 19, 1871, p. 755.
dx gS
- =u+c+ --:-c~-:---~ Boussinesq, J., Essai sur Ia theorie des eaux courantes. Institut de
dt - - (ojox)(u 2c) France, Academie des Sciences, Memoires presentes par divers
savants, 23, 1877.
18.8 Transform the differential equations which are used Broer, L. J. F., On the propagation of energy in linear conservative
in the method of characteristics for the study of the propaga- waves. Appl. Sci. Res., A2: 447-468, 1951
Chapter 18: The Long Wave Theory
Carr, J. H., and Stelzriede, M. E., Diffraction of Water Waves by Koh, R. C. Y., and Le Mehaute, B., Wave shoaling. J. Geophys.
Breakwaters in Gravity Waves. U. S. National Bureau of Res., 71: 2005-2012, 1966.
Standards, Circular 521, pp. 109-125. Kotschin, N. J., Kibei, I. A., et a!., Theoretische Hydrodynamik.
Chappelear, J. E., Shallow water waves. J. ofGeophys. Res. 67, 1962. Adad. Verlag, Berlin, 1964.
Coulson, C. A., Waves, 4th ed. Oliver and Boyd, Edinburgh, 1947. Korteweg, D. J., and de Vries, G., On the change of form of long
Dane!, Pierre, On the Limiting Clapotis. U. S. National Bureau of waves advancing in a rectangular canal on a new type of long
Standards, Gravity Waves, NBS Circular 521, 1952, pp. 35-38. stationary waves. London, Dublin and Edinburgh, Philosophical
Magazine, Ser. 5, 39: 422, 1895.
Dean, R. G., Stream Function Wave Theory-Validity and Applica-
tion. Specialty Conference on Coastal Engineering, ASCE, Kravtchenko, J., and Daubert, A., La Houle a trajectoires fermees
1965. en profondeur finie. La Houille Blanche, 12, 1957.
Druet, C., Nomographic chart for determination of the monochro- Laitone, E. V., The second approximation to cnoidal and solitary
matic wave type in the region of foundation of a designed waves, J. Fluid Mech., 9, 1960.
hydrotechnical structure. Paper S, 11-I, 21st International Lamb, H., Hydrodynamics. Dover Publications, New York, 1945;
Navigation Congress, Stockholm, 1965, pp. 183-201. Cambridge University Press, 1932.
Dubreuil-Jacotin, J. L., Sur les ondes de type permanent dans les
Le Mehaute, B., Mass transport in cnoidal waves. J. Geophys. Res.,
liquides heterogenes. Rendiconti della Accademia Nazionale dei
73: 5973-5979, 1968.
Lincei, Ser. 6,15: 814--819, 1932.
Dubreuil-Jacotin, J. L., Sur Ia determination rigoureuse des ondes Le Mehaute, B., Divoky, D., and Lin. A., Shallow water waves: A
permanentes periodiques d'ampleur finie, Journal de Mathemati- comparison of theories and experiments. Proc. Amer. Soc.
ques Pures et Appliqw!es, Ser 9, 13: 217-291, 1934. Civ. Eng. (lith Conf. Coastal Eng.) 1, 86--107, 1968.
Eckart, C., The Propagation of Gravity Waves from Deep to Le Mehaute, B., and Webb, L., Periodic gravity waves over a gentle
Shallow Water. U. S. National Bureau of Standards, Gravity slope at a third order of approximation. Proc. Amer. Soc. Civ.
Waves, NBS Circular 521, 1952, pp. 165-173. Eng. (8th Cong. Coastal Eng.) 1: 23-40, 1964.
Freeman, J. C., and Le Mehaute, B., Wave breakers on a beach Le Mehaute, B., Theory of Explosion-Generated Water Waves.
and surges on a dry bed. Proc. Amer. Soc. Civ. Eng., 90: 87-216, Advances in Hydroscience, Vol. 7, Academic Press, New
1964. York, 1971.
Friedrichs, K. 0., On the derivation of the shallow water theory,. Le Mehaute, B., Theory of wave agitation in a harbor. Trans.
Appendix to the formation of breakers and bores by J. J. Amer. Soc. Civ. Eng., 127: 369-383, 1962.
Stoker. Comm. Pure Appl. Math. I: 81-85, 1948. Longuet-Higgins, M. S., and Stewart, R. W., Radiation stress and
Gerstner, F., Theorie der Wellen, Annalen der Physik, 32, 1809. mass transport in gravity waves. J. Fluid Mech., 13: 481-509,
1962.
Goda, Y., Wave Forces on a Vertical Circular Cylinder. Report
No. 8, Port and Harbor Technical Research Inst., Japan, 1964. Longuet-Higgins, M. S., Mass transport in water waves. Phil.
Trans. Roy. Soc., A 245: 535-581, 1953.
Ippen, A. T. (editor), Estuary and Coastline Hydrodynamics.
McGraw-Hill, New York, 1966. Mei, C. C., and Le Mehaute, B., Notes on the equations of long
waves over an uneven bottom. J. Geophys. Res., 71: 393-400,
Johnson, J. W., O'Brien, M. P., and Isaacs, J. D., Graphical 1966.
Construction of Wave Refraction Diagrams, Hydrographic
McGowan, J., On the solitary wave. London, Dublin and Edinburgh,
Office, Navy Department, Publication No. 605, 1948.
Philosophical Magazine, Ser. 5, 32: 45, 1891; On the highest
Keller, J. B., The solitary wave and periodic waves in shallow water. wave of permanent type. Ser. 5, 38: 351, 1894.
Comm. Appl. Math., December 1948.
McNown, J. S., Waves and Seiches in Idealized Ports. U.S. National
Keulegan, G. H., and Patterson, G. W., Mathematical theory of Bureau of Standards, Gravity Waves, NBS. Circular 521, 1952.
Irrotational translation waves. U.S. NBS J. Res. 24, 1940.
Miche, A., Mouvements ondulatoires de Ia mer en profondeur
Kinsman, B., Wind Waves. Prentice-Hall, Englewood Cliffs, New constante ou decroissante. Annales des ponts et chaussees,
Jersey, 1965. 1944, pp. 25-78, 131-164, 270-292, 369-406.
Part 3: Water Wave Theories
272 Mitchell, J. H., The wave resistance of a ship. London, Dublin and
Edinburgh, Philosophical Magazine, Ser. 5, 45, 1898.
Milne-Thompson, L. M., Theoretical Hydrodynamics. Macmillan,
London, 1938.
Munk, W. H., The solitary wave theory and its application to surf
problems. Annals New York Acad. Sci., 51: 376-424, 1949.
Munk, W. H., and Arthur, R. S., Wave Intensity Aloi).g a Refracted
Ray. U. S. National Bureau of Standards, Gravity Waves,
NBS. Circular 521, 1952.
Pierson, W. J., Jr., The Interpretation of Crossed Orthogonals in
Wave Refraction Phenomena. Technical Memorandum No. 21,
Beach Erosion Board, Corps of Engineers.
Putnam, J. A., and Arthur, R. S., Diffraction of water waves by
breakwaters. Trans. Am. Geophys. Union, 29: 1948.
Rouse, H., Fluid Mechanics for Hydraulic Engineers. McGraw-
Hill, New York, 1938.
Stokes, G. G., On the theory of oscillatory waves. Trans. Cambridge
Philosoph. Soc., 8, 1847; and Suppl., Scientific Papers, 1.
Struik, D. J., Determination rigoureuse des ondes irrotationnelles
periodiques dans un canal a profondeur finie. Mathematische
Annalen, 95: 595--634, 1926.
Suquet, F., Remarks on graphical computation of wave refraction.
International Association for Hydraulic Research, Grenoble,
1949.
Sverdrup, H. U., A Study of Progressive Oscillatory Waves in
Water; A Summary of the Theory of Water Waves. Technical
Reports No. 1, 2, Beach Erosion Board, Office of Chief Engine-
ering, U. S. War Department.
Thomas, H. A., The Propagation of Stable Wave Configurations in
Steep Channels. Carnegie Institute ofTechnology, Pittsburgh, Pa.
Thomas, H. A., Hydraulics of Flood Movements in Rivers. Carnegie
Institute of Technology, Pittsburgh, Pa., 1937.
Thomas, H. A., Propagation of waves in steep prismatic conduits.
Proceedings of Hydraulics Conference, University. of Iowa,
Studies in Engineering, Bulletin 20, 1940.
Ursell, F., The long-wave paradox in the theory of gravity waves.
Proc. Cambridge Philosoph. Soc., 49, 1953.
VanDorn, W. G., Tsunami. Advan. Hydrosci., 3: 1--48, 1965.
Wehausen, J. V., and Laitone, E. V., Surface waves, In Handbuch
der Physik, Springer-Verlag, Berlin, 1960.
Wiegel, R. L., Oceanographical Engineering. Prentice-Hall, 1964.
Whitham, G. B., Mass momentum and energy flux in water waves.
J. Fluid Mech., 12: 135-147, 1962.
Appendix A A-1 Introduction to a Real Sea State
273
~"'""ATE,
WAVE PERIOD /'
Appendix A: Wave Motion as a Random Process
i
waves as random processes is now presented.
AI A4
A-2 Harmonic Analysis A2
I
T! -i + L -r;- -r;-
A [ 2nnt . 2nnt]
ry(t) = An COS + Bn Sill
Figure A-4
TIME A typical wave record at sea.
0
i
An = - l](t) COS--
T, 2nnt . 2nmt T1 o T1
COS - - Sill - - dt = 0 for all m and n
o T1 T1 and similarly by multiplying by sin [(2nm)/T1]
i
o
2nnt
T, 2nmt
cos-- cos-- dt = 0
T1 T1
form f= n Bn = -
2 iT, 2nmt
l](t) sin-- dt
T1 o T1
i
function IJ(t) can be represented by this set of equations
T, . 2nnt . 2nmt d cannot be stated precisely and fully within the scope of
Sill - - Sill - - t =0 form f= n
o Tl Tl this book. (The reader is referred to many standard text-
books on Fourier analysis.) It is sufficient for the present
form= n purpose to write one of the necessary conditions, which is
+ L An
n
I
T,
o
2nnt 2nmt
COS - - COS - -
T1 T1
dt
points for each coefficient will occur at multiples of kn =
(2n)/T1 where T1 is the length of the record.
In order for the Fourier coefficients An and Bn to be
+ L Bn
n
i
T1
o
2nnt 2nmt
Sill - - COS - -
T1 T1
dt
representative of a sea state for all time, the record length
T,. must be very long. The limits on n will be very large
and the practical details of such a Fourier representation
AmT1 are seen to be enormous. A comparatively short record
(n = m)
2 of the sea will yield values for An and Bn which would
Appendix A: Wave Motion as a Random Process
value at timet + !lt. The correlation function density. The energy density spectrum really expresses
the rate of change of variance or derivative of the signal
R(r) = lim - 1 fT' IJ(t)ry(t + r) dt as a function of frequency. However, the square of the
T,-oo Tl 0 amplitude is used instead of the amplitude. The density
predicts the expected (average) value of the product of spectrum corresponding to a simple sinusoidal wave (Fig.
two values of the water surface which are separated in A-2) would appear as a spike of infinite height at the fre-
time at the same place by r. To confirm that ry(t) and quency (2n)/T since the variation of the square amplitude
ry(t + r) are taken from the same record, the expression as a function of frequency is infinity. However, the spike
s:
for R(r) is called the autocorrelation function of ry(t). has by definition a finite area:
(Occasionally the cross correlation between two different
signals may be required.) 00
S(k) dk = !A 2
Some properties ofthe autocorrelation as defined by R(r)
are developed: Such a function is referred to as a Dirac delta function. A
periodic wave of arbitrary shape which has the amplitude
1. The autocorrelation function is even, i.e., R(r) =
spectrum of Fig. A-3 will show a comb of Dirac delta
R( -r), or
functions, each one of infinite height at the frequencies
kl> k 2 , k 3 , but having finite areas equal to Ai/2, A~/2,
lim - 1 JT' ry(t)ry(t + r) dt
T,-oo T1 0 A~/2, etc.
00
<l>(k)eikr dk
erty is beyond the scope of this book. It can be noted,
however, that if a process is random the correlation When r = 0,
between an observation at time t and an observation
taken at time r later would tend to become infinites-
imally small as the time r became large.
R(O) J oo
= -oo <l>(k) dk = },i~00 T
1 [T'
Jo l](t)ry(t) dt
-
= ry 2 (t)
It is seen that the total area under the curve <l>(k) is the mean
A-2.6 Autocorrelation to Energy (or
square value of ry(t). The area of any element <l>(k) dk
Variance) Density Spectrum
represents the mean square contribution of the variance
A-2.6.1 The concept of a spectrum (amplitude and phase) of ry(t) in the interval !bk. <l>(k) is the variance density
in the frequency domain can be introduced. The idea of a frequency spectrum of ry(t). It is often miscalled the energy
density spectrum is similar to the idea of a probability spectrum or the power spectrum or simply the spectrum
Appendix A: Wave Motion as a Random Process
of the process 17(t). It can be shown that the variance transform. 279
spectrum may be called the energy spectrum for deep water
linear waves. It is never a "power" spectrum as used in the <I>(k) = -1 f"'
2n _ 00
R(r) cos kr dr
above definition, but it is always a variance spectrum for
any water surface record-linear or nonlinear. and from the symmetry <I>(k) = <!>( -k) (by inspection) an
The reason for the term energy spectrum follows from alternative definition for the spectrum is
the definition,
then upon squaring and averaging the right-hand side One advantage of this definition is that the factor 1/(2n) in
of the above equation the Fourier transform is not required. It is also pointed
out that in order to transform a spectrum S(k) to the
equivalent S(f) it must be recalled that this is a density
=
spectrum and the identity S(f) df S(k) dk must be
maintained so that S(k) = S(f)/2n. Some authors have
It is now recalled that the average energy per unit surface also advocated the use of" period" spectrum defined by
area can be written as Eav = ipgH 2 , where His the wave
height (2An). A wave of height 2An and frequency kn in the S(T) dT = S(f) df
confused sea contributes !pgA; to the energy of that sea.
where f = 1/T. This has some advantage in giving a "feel"
It is seen from the above equations that pg J<I>(k) dk
for the effects of wave period but in fact there is no reason
represents the energy in the sea (for linear waves). The
why the same "feel" cannot be developed for frequency
description of a sea in terms of its variance spectrum
(for k = 2nf) once familiarity is achieved.
<I>(k) dk is a purely statistical definition. If physical meaning
The spectrum and the autocorrelation function have
such as "energy spectrum" is used, it should be born
many desirable properties when used to describe random
in mind that the "variance spectrum" and "energy
processes such as sea states. They (particularly the spectrum)
spectrum" are only equivalent (after a factor of pg is
have stable statistical properties in that sample records
inserted) for deep water small amplitude wave motions.
taken from the same sea state yield closely reproducible
All the operations described above are generally per-
spectra. One further obvious usefulness is the condensation
formed with high speed digital computers using fast fourier
of say a 30-min record of water surface into 50-100 points
transform (FFT) techniques. Detailed discussions of such
in a frequency domain. Attempts have been made to fit
procedures are beyond the scope of this book.
empirical curves to wave spectra.
A typical spectrum of sea-surface phenomena is shown
A-2.6.3 Advantage can be taken of the even property in Fig. A-5. This figure illustrates the complete range of the
of R(r) to reduce the Fourier transform to the cosine wave spectrum from infratidal effects to capillary waves.
Appendix A: Wave Motion as a Random Process
280
24 12 5 30 I 0.1
hr hr mrn ltl sec sec sec:
I
I ""~
!II
0
0
z., ~
cr c
Cw ltl a: ltl
-~~
g:a
ltl
""
0
ltl
""a: ""
z ""
ltl
""cr> Figure A-5
>-
C) ""
0
.....
i= C!l
::I
""C!l ..J
!II
a: -...~ ..J Energy spectrum at sea .
w """cr cr ltl
!i z
..J
z ~~ ..J
cr
z
a: ::I
a:
ltl
- ""
::I Ill
0
;;; ""3: >-
a:
w zO z
::I
0 cr 0
ltl cr
_..J
::10 a: 0 ..... z ... ..J ..J
..J
a: I ltl ::I a: II.
a:O :! ltl ::I )( iL
i
""""
:z:l-
0
""
ltl
..... ltl
"" cr
()
"""""
o:::E
A% 281
Figure A-6
Probability distribution offree
surface elevation.
a +
VALUE OF SURFACE ELEVATION
f) S(f) df = 11 2 = k; Ac
As
=
=
L An cos (knt -
- I
kmt + en)
An sin (knt - kmt + en)
and this definition is normal for spectral applications in In the equation for IJ(t) the assumption of a narrow range
most other fields. In other words, the total area under the of kn means that Ac and As vary very slowly with time.
spectrum is the variance of the process being studied. Now let R = (A? + A;) 1 i 2 . By the central limit theorem
of probability the values of Ac and As given above will be
A-3.3 Probability for Wave Heights
normally distributed so long as the sums are taken over
A one-dimensional wave train is assumed to be described a sufficiently large number of terms and
by the function
l](t) = L An cos (knt + en) A? = A; = 1] 2 = 1XlS(f) df
where the range of kn is distributed over the frequencies Therefore, the probability that Ac, As lies within the element
contained in the spectrum, the en's are arbitrary phase dAc dAs is given by the following probability density
positions and the An's are governed by the spectrum as a function
function of kn.
If the spectrum is spread over a narrow range of fre-
quencies such that the midfrequency is km and the actual
Appendix A: Wave Motion as a Random Process
In fact R can be considered as the wave amplitude H/2 where Tis defined as the mean apparent wave period (the
and ex can be considered as the wave phase position en. wave period will be twice the expected time between
The last equation indicates that the phase position of the successive zero crossings). The proof of the above equation
waves has a uniform distribution between zero and 2n. can be found in more advanced textbooks and is beyond
p(R) dR can be further simplified since for a sine wave the the scope of the current chapter. In a similar manner the
mean square value is equal to one-half of the square of the expected period between successive maxima and minima
amplitude. (points of zero gradient, OIJ/Ot = 0) is found to be given by
the ratio !(M 2 /M 4 ) 11 z, and this is seen to be quite different
Therefore, if H = 2R, IJz = !Rz =kHz so that
from T/2 for a general spectrum shape S(f). The two are
identical if S(f) is considered as a delta function having
p(H) dH = 2H e- H2fH2 dH
Hz only one frequency.
The relationship of T to the autocorrelation function is
which is the well known Rayleigh distribution. best illustrated by recalling a well-known statistical
In spite of the many apparent assumptions in the theorem. "The moment of a probability distribution of
derivation of this equation, it has been found to be an order n is given by the nth derivative of the characteristic
extremely good fit for observed wave height distributions function of the probability density at the origin." The
in wind generated seas. characteristic function of a probability density is its Fourier
Appendix A: Wave Motion as a Random Process
Figure A-7
NON- LINEAR (SKEWED)
An illustration of nonlinear effects on the
probability distribution offree surface
elevation.
-- .,..,
The effect of skewness in water surface ordinate prob- to separate the harmonics since they travel at the phase
ability distributions is reflected to some extent in the speed of the "fundamental" wave, whereas the shorter
probability distributions of wave heights. However this waves travel much slower.
effect is not very marked and does not appear to be very The spectrum destroys all details of phase position and
important. It must be recalled that as the crests become also all differences between "crests" and "troughs." The
more peaked the troughs become flatter. As far as wave normal wave spectrum is obviously an insufficient de-
heights are concerned these effects are compensatory and scriptor of a high (nonlinear) sea state.
wave height probabilities for high seas are still very close One of the major advantages of the spectral description
to the Rayleigh distribution. of a sea state became apparent in the section on subsurface
The effects on wave period distributions of large seas velocities and accelerations. The velocity, acceleration,
has not received much study. In random seas the concept pressure, etc. spectra are easily determined from the water
of"wave period" does not really exist. The only thing which surface spectrum. The transfer functions for nonlinear
may be mentioned is that the times between successive sea states will not be so simple. One other form of non-
up-down crossings of the mean water level will be shorter linearity arises in operations on the spectrum. For example,
than the times between successive down-up crossings. The the prediction of wave force spectra from sea surface
estimates of average time between successive zero crossings spectra when the function is of the form f = au 2 + b(oujot)
and successive maxima and minima as presented in the is one involving a nonlinear operation on the water surface
section on wave period probability will not be very reliable. spectrum. These problems are not yet solved satisfactorily.
A.3 If f(x) = x 2 and p(x) = [1/(2n) 1 i 2 ] exp -{!x 2 ), deter- Jenkins, G. M., and Watts, G. W., Spectral Analysis and Its Applica-
mine p[f(x)] and sketch. tions. Holden-Day, San Francisco, Calif., 1969.
Kinsman, B., Wind Waves. Prentice-Hall, Inc., Englewood-Cliffs,
A.4 Show that the autocorrelation function of any stationary New Jersey, 1965.
random variable is even. Lee, Y. W., Statistical Theory of Communication, Wiley, 1961.
National Academy of Sciences, Ocean wave spectra. Proceedings
A.5 In the determination of the spectrum of a random of a Conference, Basten, Maryland, May 1-4, Prentice-Hall,
process, the autocorrelation is first computed followed Englewood-Cliffs, New Jersey, 1961.
by a Fourier transformation in the form of a cosine trans-
formation. What would you expect to find if a sine trans-
form on the autocorrelation was performed? Why?
A.6 A random variable x has the exponential probability
density
p(x) =a exp ( -blxl)
where a and b are constants. Determine the relationship
between a and b and the probability distribution function
P(x). Sketch p(x) and P(x).
p(H) dH = H exp (
2H 2 H2
H2) dH
Establish the relationships between the most probable
wave height, the average wave height, the significant
wave height (the mean of the highest one-third of the wave),
and the highest wave (probability 0.01).
effects which are of interest in the research. A knowledge concept of a model without a priori having an idea of the 287
and understanding of the physical laws which govern the causes and origin of the phenomenon studied. The overall
effects studied permits us to determine the relative influence examination of the phenomena involved enables us, a
of" scale effects." priori, to eliminate basically erroneous methods of ap-
proaching the problem. It renders it possible to submit a
broad outline of the boundaries of the rational conditions.
B-1.2 Economic Considerations In many cases, it also will enable us to determine the princi-
pal stages of the study and occasionally, the number and
The scale is chosen as a compromise between economics type of the models required. The detailed study of the
on the one hand and the technical requirements for simili- equations ofthe phenomena will then define the approxima-
tude on the other hand. From the economical viewpoint, tions, scales, distortions, etc., and eventually, the number of
the cost of scale model experiments increases approxi- models required, as well as the rules of similitude to be
mately as the cube of the scale A3 , since the area of the model applied to each of them. For example, in the case of a
increases as A2 , and the theoretical duration of the test conventional study of the protection of a port against
increases as A112 , but the actual duration increases faster. wave action, it may happen that the installations as a
Of course, this rule of thumb has many exceptions and is whole are studied on a scale of 1/150, some particularly
not valid if the scale of the model is too small. (In the case important structures (such as those at the entrance) on a
of a very small model, the cost of the operation actually scale of 1/75, and the stability of the individual structures
increases as the difficulty of carrying out reliable measure- on a scale of 1/40.
ments increases.) The "best" scale is not the largest possible In summary, the problem of similitude is far from simple,
one. As a matter of fact, the scales of models must primarily and there are no standard solutions. The engineer has at his
be dictated by economical considerations. Actually, the disposal a number of complex means which he must be
decisive elements rarely present themselves in a precise able to handle with skill. The better he is acquainted with
manner. It usually is almost impossible to evaluate, even the tools, the more effective and economical the solution
roughly, the "economic expectation" of a model and will be. From the technical viewpoint, the problem consists
particularly the variation in this economy with respect to of examining when scale effects become important. This
the accuracy of the tests. Hence, the determination of the leads us to consider the forces which always need to be in
scales of optimum reduction will necessarily be an art similitude. Each separate case needs to be examined in
rather than a science, but the value of the results obtained detail. However, some general rules can be presented.
will nevertheless primarily depend upon the overall know- For all practical purposes, a scale model must fulfill the
ledge available to the designer. Among other things, it is following conditions:
imperative that he be acquainted with the laws of similitude
because not only will they enable him to estimate the degree 1. It must be exact; i.e., it must reproduce with exactness
of precision that he may expect from the model, but, above the natural phenomenon under study.
all, they will render it possible to have a sound concept of 2. It must be consistent; i.e., it must always give the same
the latter and to achieve the optimum precision compatible results under the same conditions.
with a given sum available for research. 3. It must be sensitive, or more exactly, its sensitivity has
The first stage of the study of similitude is not concerned to be imposed by the needs of the reproduction of the
with a detailed discussion of the equations, but is essentially phenomenon under investigation.
an examination ofthe problem as a whole and of its physical 4. It must be economical, of reasonable size, and
character. For example, it is impossible to form a sound completed within a reasonable time interval.
Appendix B: Similitude and Scale Model Technology
for example, a wave agitation pattern or the same bottom (p refers to prototype, m to model) one deduces
evolution as observed in the prototype. This faithful
reproduction of bottom evolution determines the choice of ( Vm)2
vp
= Lm =A
Lp
the wave characteristics, rather than the strict conditions
of similitude of wave motion. where A is the geometric scale. Thus the ratio of scale veloc-
Other criteria of similitude will be to what extent he ities is A112 . The ratio for time scale is also A112 , since
wants to reproduce the fineness of the bottom topography, a
typical tide cycle, a typical flood wave, the currents and Tm
their variations with time, and so on. In summary, the Tp
criteria of similitude are specified by the experimenter as The ratio of discharges is like the ratio of areas A2 times
reasonable approximations for simplification of model ratio of velocities A112 ; i.e., A5 12 . The ratio of powers is
operation. discharge times height, i.e., A7/2. The ratio of angular
velocities or frequencies 1/T is A- 112 , etc. Since the ratio of
B-2 Short Model vs Long Model pg is unity, the ratio of pressure is A, and the ratio or scale
of forces is (area x pressure) A3 . The ratio of powers
B-2.1 Froude Similitude: Nondissipative defined by force times velocity is A712 . The ratio of shearing
Short Models stress r is also A like the pressure.
B-2.1.1 There are two kinds of scale models for which the
law of similitude are deduced: short models and long B-2 .2 Generalized Froude Similitude:
models. The short models are considered either as non- Fully Turbulent Short Models
dissipative or as fully turbulent. The nondissipative short
B-2.2.1 Let us now consider the viscous forces. It is
models are the ones where the flow pattern is essentially
governed by inertia force and gravity only in the case of free known that the ratio of inertial force to viscous force is
surface flow, or inertia forces and pressure gradient only dimensionally represented by a Reynolds number, VL/v,
in the case of flow under pressure. where vis the kinematic viscosity. An equality of Reynolds
The ratio of inertial force (dimensionally equal to p V 2 ), number and Froude number is possible only at scale unity.
to gravity or pressure force (dimensionally equal to pgL) is Hence, Froude similitude is possible in two cases only:
a Froude number V 2 j(gL), which is to be the same on a (1) where viscous forces are negligible, as in a gravity
scale model as on a prototype. The Froude similitude stems wave prior to breaking or in a flow over a weir; or (2) where
from the fact that the gravity acceleration is the same on the flow is very turbulent and the flow pattern to be repro-
the model as in the prototype. An example of such a scale duced on scale model is short, as in a hydraulic jump or a
model is the flow over a spillway, when boundary layer breaking wave. Indeed, in the latter case, the dissipation of
effects are negligible. energy is mostly due to turbulent fluctuations and is not
The inertial force and the gravity and pressure forces are due to laminar viscous effect. While these viscous effects are
always present in water wave problems, and the Froude linearly related to the velocity, the turbulent fluctuations
similitude applies when viscous effects are negligible. are quadratic, i.e., proportional to the square of the average
velocity V 2 , as are the inertial forces. Thus, the ratio of dissi-
B-2.1.2 From an equality of Froude numbers, pative forces to gravity force in a very turbulent flow is also
a Froude number. This situation allows us to use the
"generalized Froude similitude." Of course, the depth at
vz'
2
gL
y 'P = gL m which the air bubble penetrates in a breaker will be relatively
Appendix B: Similitude and Scale Model Technology
larger in the field than in the model, as the size of the bubble which is in accordance with the Froude similitude, since
(determined by capillary effects) is approximately the the rate of energy dissipation is a power dimension.
same in any air entrainment phenomenon. But the dissipa- So a small hydraulic jump, defined by h 1 , h2 , and Q, is in
tive forces are in similitude and proportional to V 2 , and the similitude with a large jump defined by h 1 = Ah'1 , h2 =
total amount of energy dissipated is in similitude (A 7 12 ), Ah~, and Q = A512 Q' at a scale A. Indeed, the phenomenon is
even though the fine structure of the flow could be different. very turbulent. The average flow pattern is the same. The
turbulent fluctuation may statistically be different, but the
B-2.2.2 This is evidenced when one considers, for gross effect remains the same. It is the reason why stilling
example, the shearing stress (see Section 8-3.3): basins, for example, could be investigated by scale models.
In the case of breaking waves, it is sufficient to remember
- au (au) 2 au that scale effects will be negligible for the phenomena under
r = pu'v' + 11 ay = p/2 ay + 11 ay consideration, provided the height of breaker is larger than
where uis an average velocity, 11 the viscosity, p the density, say, 5 em. This statement is based more on experience than
l the Prandtl mixing length, and u' and v' the turbulent on theory.
velocity components. In the case where rm = A:r P and
a- ~ p [2
11 __!!_
ay
(a-)__!!_
ay
2
B-2.2.4 From the most general viewpoint, it is known
that the average fluid flow obeys the Reynolds equation
au; -au; a(p + pgz) a 2 U; a ( -,-,)
it can easily be seen that the shearing forces r will be in p-a
t
+ puj-a
xi
= a
X;
+ 11 aX;- a
xi
--a pu;uj
xi
similitude provided the mixing lengths are such that l' = AI
in accordance with the similitude of Froude. This condition where (U;) is the average velocity and u; is the turbulent
is fulfilled in the case of fully turbulent motion occurring fluctuation velocity. It is known that u; = u; and u; = 0 in
over a short distance and presenting a large velocity the case of a perfect fluid (11 = 0) and in the case of a laminar
gradient. flow. If this equation applies for the prototype, a similar
equation should also apply for the scale model, and for
similitude one should have
B-2.2.3 An example will illustrate these considerations.
The rate of energy dissipated by a hydraulic jump or a _au; I
pui- -I
a(pgz)
-
moving bore, dE/dt, depends only upon the discharge Q axi m ax; m pu;u}lm
and the depth of water before and after the jump (h 1 and h2 ). -au; I = ... = a(pgz)l - pu;ujlp
One has seen (Section 14-5.1.2) that the rate of energy dissi- pui- --
axi p ax; p
pation by an hydraulic jump is:
dE _ Q (h 1 - h2 ) 3
=
Since Pm = pP, g g and zm/zP = A, one may deduce that
the scale for pressure p is also A. The scale for acceleration
dt - pg 4h1h2
au;/at is unity, as is g. Consequently, the scale for time is
independently of the scale. Therefore, the ratio A112 , and the scale for velocity is
is possible only if u;uj :::::; (~) 2 which is the case of very the inertial forces. This is the generalized Froude similitude 291
turbulent flow on short structures. It is also seen that seen in the previous section.
similitude is possible, as previously stated, in the case of a On the other hand, in a long model, friction has a
perfect fluid (f.l = 0, il; = u;), where the only forces are definite influence on the flow pattern; therefore, in addition,
inertia (linear or nonlinear) and gravity-pressure. Con- a similitude of head loss is required. This head loss is a
sequently, a first scale effect is due mostly to the fact that the function of the Reynolds number, but is not determined by
viscous forces are not in similitude. The value of scale model the so-called Reynolds similitude requiring an equality of
experiments depends upon their relative importance, i.e., Reynolds numbers. Therefore, similitude for long models
the thickness of the boundary layer and its effect on related requires, in addition to the Froude similitude, another
phenomena such as separation and wakes. Fortunately, condition which makes the similitude sometimes impossible
many fluid flows may be successfully considered as that as it is shown in the following section.
of a perfect fluid (f.l = 0) or as very turbulent. The Froude
similitude or generalized Froude similitude is then the rule B-2.3.2 Indeed, the head loss /1H in a pipeline is
for all short flow patterns.
11H _ (UD) U 2
B-2.2.5 In the case of flow under pressure, which could l - f v 2gD
be subjected to cavitation the similitude parameter where lis the length of the pipe, Dis its diameter, and U is
11pj(p V 2 ), may intervene requiring to reduce the atmospheric the average velocity. For similitude of inertia and gravity,
pressure Pa to a scale model partial vacuum. one must have
B-2.3 Similitude of Head Loss; Long
Models
uzl uzl
gD m = gD P
possible to adjust by trial and error the scale model rough- B-2.3.4 Ship resistance is due to the waves created by
ness in order to obtain the same friction coefficient. A the hull displacement, and to the boundary layer drag and
"Froudian" discharge is then obtained, i.e., Qm = QPA 5 12 , flow separation. The first phenomenon-the ship wave or
when the relative roughness Kelvin wave-would exist even in a perfect fluid, and it is a
function of a Froude number only. From this standpoint,
the ship is a short model and this effect could be investigated
in similitude. But the boundary layer development along
the hull is a long model, and since the hull is smooth, this
are theoretically identical, and the Reynolds number force is not reproduced in similitude. One can only obtain
(U Djv) lm exceeds a given value, in such a way that the an approximate effect, introducing a thin wire at the bow
boundary layer is "completely rough," or again k 5 U*jv of the scale model, in order to cause some turbulence in the
> 70 (see Fig. 13-15). But, in any case, the velocity distri- scale model boundary layer. Similarly, waves breaking on
bution which is also a function of the Reynolds number the large armor blocks covering a breakwater core create
(see Fig. 13-13) could never be in similitude exactly. It is a very violent and turbulent flow motion around these
scale effect, which is most often acceptable. The same blocks. Therefore, it is to be expected that the corresponding
considerations apply to free surface flows. pressure forces will be in similitude, and studies on stability
of breakwater are done according to the law of the general-
B-2.3.3 Water wave scale models could, in general, be ized Froude similitude (a short model). The study of the
considered as short models, whether in Froude similitude permeability of breakwater requires the reproduction of
(wave propagation) or in the generalized Froude similitude similitude of permeability and energy loss by distorting
(wave breaking). The magnitude of long shore currents and the size of the material of the core of the breakwater. From
location of rip currents may, to some extent, depend upon this standpoint, these are long models, since the stream
friction characteristics of the beaches, in which case the tubes within the porous core are long.
study of this phenomena would have to be considered as
belonging to the categories of long models and, therefore, B-2.4 On Distortion
may not be studied on scale models. Nevertheless, most
scale model studies have to deal with short coastal structures B-2.4.1 A model is distorted when one of its scales (say
(like entrance of harbor); thus, the water motion is not too the vertical scale Jl) is different from the horizontal scale A.
dependent upon the friction coefficients. The main dissipa- D = 11/A is the rate of distortion. A long model can also
tive mechanism is due to wave breaking. If viscous damping have two vertical scales, one for cross sections and another
is too significant, as is the case when the wave has to travel one for the bottom or energy slope, For example, the slope
a long distance in very shallow water, a correction co- of a scale model river could be adjusted to satisfy a similitude
efficient, determined theoretically, can be applied. Very condition for head losses.
rarely do we have to be concerned with adjustments of
roughness for similitude of energy dissipation, and coastal B-2.4.2 A short model cannot be distorted as the flow
models can be considered as short models. On the other pattern on the scale model would be completely different
hand, as has been pointed out previously, a similitude of from what it is on the prototype. A long model could be
head loss is imperative for models of rivers and estuaries- distorted to some extent, without too much scale effect. It is
these are long models. Despite this adjustment, the vertical actually a method frequently used to overcome the con-
velocity distribution which is a function of the Reynolds straint imposed by the required equality of friction co-
number can never be in similitude. efficient between model and prototype. Also, it permits
Appendix B: Similitude and Scale Model Technology
reduction of the horizontal scale of the model (i.e., the cost) B-2.4.5 In the case of a movable bed scale model which is 293
without sacrificing too much accuracy. used to study sediment transport and bottom evolution of
rivers, estuaries, and beaches, distortion is not an engineer-
B-2.4.3 In gen~ral, scale models used in the study of ing trick for reducing the size of the model and the bottom
water waves cannot be distorted. The wave velocity is friction, but is the extrapolation of a natural observed
phenomenon. The method to obtain a satisfactory scale
gT 2nd model is first to obey the law of nature, even though this law
C =-tanh~
2n L may not be fully understood. For example, a small river
The tanh lm must be the same as the tanh IP and this is only flowing in its own alluvium can be considered a distorted
possible if model of a large river. This means that the ratio depth to
width of the small river is comparatively greater than the
relative depth of the larger one. The ratios of depths J1 and
widths A are approximately related by the law of Lacey
(A 2 = J1 3 ), in accordance with the" regime theory." Similarly,
Then both time scale and wave velocity scale are like A112 a beach in a protected area has a relatively steep slope,
Since the ratio of wavelengths L,JLP is given by the hori- whereas a beach in an exposed area tends to have a more
zontal scale A, it is deduced that the ratio of water depth gentle slope. The vertical scale being defined by the ratio of
dm/dP should also be A. Then the wave pattern is in similitude. incident wave heights, a protected beach can be considered
As such, water wave models could be considered as short as a distorted scale model of an exposed beach. In both the
models despite some viscous damping. There are some case of the river and the beach, the choice of distortion
exceptions in the case of long waves and in the case of becomes a stringent condition to be respected quantitatively.
movable bed scale models. However, the natural law determined by statistical observa-
tion of natural phenomena needs to be modified in the case
B-2.4.4 In the case of long waves of a river model because scale models generally do not use
2nd the same material as the prototypes in order to satisfy
h 2nd
tan~~~ other conditions of similitude. In the case of beaches, a
L L
natural distortion based on the use of sand both in the
and scale model and in the prototype is unavoidable. However,
c= (gd)1f2
the use of sand is not always recommended because it leads
L = T(gd)1f2
to too large a distortion and, subsequently, to large scale
The model could then be distorted. The velocity scale is effects. Lighter materials (pumice, coals, plastic) are
preferred.
294 exact similitude of water motion. Since the model is After some straightforward calculation, one obtains the
distorted, similitudes of wave refraction and wave breaking scale effect defined by 11L/L from the following equations:
are only aimed at. These conditions are the most susceptible In the case where d/L is small,
to producing satisfactory long shore current and sediment
!1L 1.55 x w- 3
transport distribution. This is achieved by keeping the
L~ dT 2
ratio of wave lengths and wave heights like the vertical scale
Jl. Based upon this condition, the following wave character- where dis in centimeters and T is in seconds. If (2nd)/L is
istics are preserved in the model: (1) wave steepness; large,
(2) refraction pattern and angle of refraction with bottom
!1L ys
contours; (3) breaking angle of wave crests with shorelines, if
the distortion is not too large; and (4) breaking depth. T ~ (0.105)4
Also, the scale for long shore current and mass transport Figure B-1 illustrates these results, where 11L/L is given as
velocities is approximately J1 1 / 2 . Therefore, the ratio of a function of T and d. It is seen that 11L/L is smaller than
scales of wave particle velocity to current velocity is 1 %when T > 0.35 sec and d > 2 em. In practice, it can be
approximately unity. The study of movable scale model stated that the lower limit of limiting scale effects is
technology is beyond the scope of this book. T > 0.35 sec.
2n p L' L'
where L' is the wave length (C = L'T), T the wave period,
g the gravity acceleration, d the water depth, p the density,
and A the surface tension (74 dynjcm for air-water interface
at 20C). If L is the wavelength when A is assumed to be
equal to 0, then
PERIOD T, sec
Appendix B: Similitude and Scale Model Technology
"' ~ (
where L\ is the damping coefficient.
In a rectangular tank of width b and depth d gp)~'
Pb;. Ps K!i cot IJ.
L\ = L\b + L\s
L\b is due to the solid walls and is as presently used by engineers. Ka is a damage coefficient
~ 3 for rocks. Therefore, for similitude:
L\b =2m (~) 112 mb + sinh 2md
b 2k 2md + sinh 2md
where m is the wavenumber and equals 2n/L, k is the
frequency and equals (2n)/T, and vis the kinematic viscosity.
L\ 5 is due to a contaminated film on the free surface which,
at the maximum, can be where p is the fresh water density. It is found that
L\
s
= 2m (~) 112 mb cosh 2 md
b 2k 2md + sinh 2md
(p,,:~ p')' ~ (p'~~ p)'
From these formulas, it can be verified that the viscous If Pbm = Pbp it is seen that the error on W due to the use of
damping is not negligible in many wave tanks where the fresh water is about 10%.
water depth is smaller than 2 em. This water depth should B-3.4 Wave Forces on Structures
be considered as a reasonable lower limit for scale model
studies. The important case of a cylindrical structure is now
analyzed. One must distinguish between large piles and
small piles, characterized by the value of the ratio of the
B-3.3 Density Effects diameter D to the wavelength L: D/L. In the first case
(small D/L), the wave force is given by the Morison formula
One important scale effect is due to the fact that in the as a function of the horizontal component of velocity u
wave tank one usually uses fresh water instead of sea water. of the incident wave as (see Section 13-5.2.8):
The slight difference in density, which is approximately 3 %,
changes the wave forces accordingly. Let us consider, for
example, the case of a rockfill breakwater. The minimum dF = (PcvD u; + pCM n~ 2 ~~) dz
Appendix B: Similitude and Scale Model Technology
where ev is a drag coefficient and eM is the virtual mass at a small scale. Moreover, complex boundary conditions
coefficient. ev and eM are supposed to be constant param- due to complex structural forms and nonlinear effects,
eters which are functions of the Reynolds number, (uD)jv, particularly important in the case where the wave breaks
roughness of the pile, etc. Actually, ev and eM are time on the structure, are also in similitude, whereas they cannot
dependent functions (but these functions are still not well be analyzed by theory.
defined), and the wave velocity field u is not free, but All these considerations being borne in mind, it is pos-
modified by the presence of the pile. The coefficients ev sible to establish the critical value of H/D for which inertial
and eM are obtained experimentally. The coefficients eM forces
and ev must be the same on the model and on the proto-
type. For a perfect fluid, eM = 2 and ev = 0. In practice, a (F M)max = peM nD 2
-
4
J~ (au)
;;- dz
wake develops on the lee side and on the front side of the -d ut max
pile alternately, hence eM # 2 and ev > 0.
The wake effect does not change the value of eM very equal the drag force
much, but it strongly affects the drag coefficient ev. The
dependence of ev upon the Reynolds number makes simili- (F v)max = pevD J~ U~ax dz
-d
tude possible only if (uD)/v > 2 x 10 5 in the case of
uniform flow, but actually u is time dependent and varies as a function of the wave height H and the relative depth
from positive value to negative value alternately. The effect D/L. u, at a first order of approximation, is equal to:
of roughness remains small, except that it may change the
inception of flow separation and hence influence the wake. _ k H cosh m(d + z) (k _ )
From this standpoint, the development of the boundary u - 2 . h d cos t mx
sm m
layer around the pile is to be considered as a long model
(even though the body itself is short). Therefore, wave The calculations have been performed for two values of the
forces on small piles cannot be studied accurately on scale wave height, namely H ~ 0 and H = Hb, such that Hb/Lb =
models unless the ratio of inertial force to drag force is 0.14 tanh (2nd/Lb) (limit wave steepness).
large. This ratio is represented by the Iversen modulus For the sake of simplicity, this work has been carried
out by assuming ev = 1 and eM= 2 as the most realistic
aul D
average values. The results of these calculations are
presented in Fig. B-2. On the high side of these two curves,
ot max
the drag forces dominate and there is no way of obtaining a
U 2 1max
satisfying scale model investigation except at scale unity.
which is actually an inverse Froude number where the On the low side, the inertial force dominates in such a way
gravity acceleration g is replaced by the particle accelera- that similitude is more valid.
tion oujot. From these considerations, it is seen that in the case of
In the case of very large piles (large D/L) the problem is small D/L, the larger the scale, the better. However, large
one of wave diffraction. The drag force is small, and in some values of D/L can be studied satisfactorily even at a small
cases (like large circular or ellipsoid cylinders), the problem scale.
can be linearized and treated completely analytically with This nomograph can also be used as a guide for forms of
good accuracy. The solution is given by a potential function. body other than circular pile. For example, if D is con-
It is evident that such wave motions around very large piles sidered as the beam of a ship, it is seen that the study of ship
(D/L > 0.5) are very well reproduced on scale models even behavior in water waves can be done in similitude.
Appendix B: Similitude and Scale Model Technology
prototype observations.
6. Shock waves due to underwater explosions are not in F(s) is a function of the void coefficient which will be
similitude since the compressibility of water is not assumed to be the same on the model as on the prototype.
scalable. The same reasoning applies to slamming or
impact phenomena. The free surface is "harder" on
scale models.
REFERENCES FOR APPENDIX B
7. The wave reflection coefficient of a smooth wall is
smaller on a scale model than on the prototype. The Biesel, F. and Le Mehaute, Note on the similitude of scale models for
opposite tends to apply in the case of wave absorbers studying seiche in harbors. La Houille Blanche, 3: 392-407,
made of ripraps (large rocks). This scale effect is July 1955.
corrected by adding a wire mesh on the scale model Birkhoff, G., Hydrodynamics. Princeton University Press, Princeton,
wave absorber. New Jersey, 1950.
Appendix B: Similitude and Scale Model Technology
(LY
Also coefficient of angular deformation
~0
UR=y d Ursell parameter
h* = !!_ Reduced water depth
E
Uo Velocity outside the boundary layer
he Critical depth
v Velocity vector
hn Normal depth
V(ii,B,w) Average velocity vector with respect to ht> h2 Conjugate water depths
time and its components in a turbulent
flow k Coefficient of permeability for porous
medium
V(u,v,w) Average velocity vector with respect to Also coefficient of von Karman in the
space and its components in a flow theory of turbulence, universal constant
through porous medium ~OA
V'(u',v',w') Velocity vector for the turbulent fluctua- Also 2n/T in the wave theory and periodic
tions and its components motions where T is the period
Notation
k = cp CP is the specific head at constant pressure <l>m Dissipation function due to viscous force 303
cv and C" the specific heat at constant volume
<I>, Dissipation function due to turbulent
ks Characteristic number for roughness size in fluctuations
pipes and wall boundaries (X Correction factor for the kinetic energy
Prandtl's mixing length term in a pipe
Also width of a channel at the free surface Also (tan rx) slope
m Mass; 2n/L rx,{J Constant parameters in the Gerstner wave
n (or An) Element perpendicular to an area or to a theory
streamline b Diameter of a particle of a porous medium
Also element of an equipotential line Also boundary layer thickness (general
p Pressure; probability density function definition)
Ap Pressure difference over a finite interval b* Displacement thickness for a boundary
layer
p* p + pgz: pressure and gravity force
e Coefficient of Boussinesq for shearing stress
p Average pressure with respect to time in a due to turbulent exchange
turbulent flow
Also void coefficient for a porous medium
p Average pressure with respect to space in a
flow through porous medium IJ Elevation of the free surface around the still
water level
q* = g/E(2gE)ll2 Reduced discharge
Also rotation in (x, z)
ro Radius of a pipe
fj Average elevation of the free surface
kD around still water level or wave set-down,
s=- Strouhal number
v set-up
t, T Time
u Local velocity in a pipe or in a boundary 8 tan 8: bottom slope
layer Also momentum thickness for a boundary
u* = (ro/P)l/2 Shear velocity layer
Also longitude in a spherical system of
u, v, w Components of the velocity vector V along coordinates
the three coordinate axes OX, OY, and OZ,
respectively .A Second coefficient of viscosity for gas
Also scale, horizontal
v, Component of the velocity vector along a
radius in a cylindrical system of coordinates Coefficient of viscosity
fJ
Vo Component of the velocity vector Also scale, vertical
perpendicular to a radius in a cylindrical
system of coordinates v Kinematic coefficient of viscosity (v = tJ! p)
x,y,z Coordinates of a point along OX, 0 Y, and ~ = ~cw-
2 ay
av)
az
OZ, respectively
r Circulation of velocity IJ = ~ cu _ aw)
Coefficients of rotation along 0 X, 0 Y, and
OZ, respectively
2 az ax
<I> Dissipation function
Also angle of latitude in spherical ( = ~ cv _au)
coordinates 2 ax ay
Notation
r Shearing stress 1 82 A oA
2
to Selected Y - Yo = - - - (x -
c
A+ Bt 0
x0)
Problen1s Paths:
X - Xo =
A
C (y - Yo) +l
1 B
C 2 (y - Yo)
2
{
x - x0 = Roc + r sin a
z - z0 = R + r cos a
which are the parametric equations of a trochoid.
1.5 Streamlines are
K
~- = sinh m(d + z)
cos mx
where K is a constant of the integration.
Paths: Calculate (x- x,) 2 , (z- z,f and add. Solution:
(ellipse)
where
H cosh m(d +z
0)
A = - -----:---:-----c~
2 sinh md
H sinh m(d + z0 )
B = - ---:----:-----:--
2 sinh md
F = y3 - (x + Ut) =0
Boundary condition:
- U - u + 3vy 2 =0
Answers to Selected Problems
1.10 The motion is steady with respect to the considered system co- i3p 1 i3 1 i3(pve) i3(pvz)
3.3 - + - - (prv) + - - - + - - = 0
ordinates. Then relative streamlines and paths are identical and defined i3t r or ' r ae i3z
by sinusoids of horizontal axis. The length L (m = 2n/L), and the ampli-
tudes are 3.4
H sinh m(d + z)
in the case defined in Problem 1.5 3.7 1 i3 2 1 i3 . 1 OVq,
2 sinh md 2 -(v,r) + - - - ( v6 sm 0) + - - - = 0
r or r sin e ae r sin e i3<1>
The free surface and bottom are streamlines.
-1-i3 ( r
2 i3<P)
- +- 1 -- i3 ( sme-
. iJ) + - -1 - -iJ2
-=0
r2 or or r 2 sine ae ae r2 sin 2 e iJ<I> 2
2.1
au 0.1/t
nD 2 i3u
--dxdt
-
ox = 4 ox
2.3 a = 0, h = - ( = 5/t; all other coefficients are zero. and is equal to the sum of the following:
1. The change in volume by compression is:
2.4
a = 0, h = - ( = ~ [o:y - _:_ (o: - 2Jl2v)] nD 2 i3p dx
2 Jl 2)l e --dt-
4 i3t K
If o: = 0, V oft 0 the flow is created by the moving plane; if o: oft 0, V = 0
the flow is due to a gradient of pressure dp/dx. 2. Since the increase in the stress in the pipe wall is equal to
The derivative of the volume due to variation of free surface elevation is 4.9 r-direction:
a1J
- (a 2 sin ci> ae aci>) av, av, Voav, av, V~
at A =-+v - + - - + v - - -
' at ' ar r all z az r
The discharge through two adjacent side elements are:
UEhasinci>a8 and U8 haaci> 11-direction:
Chapter 4
av a(V2 /2)
4.1
dV
- = 1.72 L/t 0
2 4.1o 1. - + -a- and
dt at x
4.2 dT n ret
-=-cos--- oF /hr, where t is in hours. au a(V
2.-+--,
2 /2) aw
dt 3 12 40 at ax at
(If oF and miles are replaced by oc and kilometers, respectively, then one
finds the same solution in C/hr.) _2w_c_o_s_C!>_U_E = 7.65 x 10 _ 6
4.12
141 -
4.4 I. 1J = [ - 3-2 -
t] 2
ft = [0.172(141 - t)] 2 em (tin sec)
g
au Chapter 5
2. p - = -0.16lb/ft 3 = -0.00256 gjcm 3
at
au
3. pu- = 1.78rc(2g1}) 112 lb/ft 3 = 0.0284n(2g1]) 1 i 2 gjcm 3
5.1 -2)1(()~
az
- axa'), _ 2Jl(a1J _a~)
ax ay
ax
4.7 11-direction:
Wmax = ka at X = 0, l, and Z = d
z-direction:
a
-
d
Answers to Selected Problems
308 6.10
5.5
:: = ~fr(<~)
u = pgj (R 2 - r 2)
4J1,
u(z) = 1 e 2 ) - 1 ][e 1
U { 1 - [ 1 - ( 1 + J.iJ.i-- 2 ~ + ee2 - z]}
when z > e1 au I
J.i,- ov
=OJ.i,- I =-r (u, v-+ 0 when z -+oo)
2 OZ z~O ' OZ z~o
2. u = (pgjj2J.i,)(h 2 - z 2 ), Q = (2pgj/3J1,)h 3 which gives, with a= (pw sin <ll/J.i-) 1 12 and U 0 = r/J.i-a(2) 1 i 2
3. ii = pgjh /3J.i-, u = (3iij2h 2 )(h 2 - z 2 )
2
4. d2u/dz 2 = -3ii/h 2, dp/dx = -3J1,ii/h 2 u= U 0 e-az cos(~- az)
5. ~ = 0, ' = 0, Yf = 1{aujiJz) =- pgjz/2J.i-
6. pgjQ = 2(pg) 2jW /3J.i-, j = 3J.i-/2pgh 3 It is also found that
7. w = 0
1 iJp 2 iJ2u
--=vVu~v-
p iJx iJz 2 At z = 0 IV I = U 0 and is at 45 angle with the wind stress (on the right
1 iJp 2 i]2u in the northern hemisphere and on the left in the southern hemisphere).
--=vVv~v The projection ofthe velocity vector is like a logarithmic spiral p = U 0 e- 9 ,
p iJy iJz2
and the angle (} of the velocity vector with the axis OX is (} = rr/4 - az,
iJpjox = -3J1,ii/h 2 , opjoy =- 3J.i-v/h 2 (ii = v average), oiijox + oiijoy = 0. i.e., varies linearly with depth.
Since o2pjox oy = o2 p/oy ox, one must have oii/oy - oii/ox = 0. </> =
-h 2p/3J.i-.
Chapter 9
6.9
9.1 One layer:
llH V 2 (1 - ef 24v
-=C - - - C =-
IlL x 2gfl e2 ' x Vfl
Answers to Selected Problems
Chapter 10
and
o 2 ifr o2 1/1 ol/1 oifr o 2 1/1 ) op* 0 2
(
+ ox ox oy = - ay - Jl ox v "'
10.2 2(u( - w~) = - -
oy 2
(V
0 - + -p + gz )
P
2
p - ox ot- ox 2 oy
2(v~ - u17) = - -
0
oz
(V2
-
2
+ -p + gz )
p
11.5 1/1 = Uy[l- ~]
x2 + yz
= u(r- 2
R )sin0
r
10.3 v, = 0, Ve = -l.SU sine, p - Poo = !pU 2 (1 - sin 2 tl) 11.6 u = 0; v = 2x; oufoy - ovfox =- 2 (rotational); vorticity 2( =- 2;
oujox + ovjoy = 0 (incompressible). No equipotential line; it is a Couette
10.4 The equation of streamlines yields the following equalities: flow between two parallel plates.
v dx = u dy, w dx = u dz, w dy = v dz 11.7"' = h2
which gives an expression such as u(oujox) dx = u du. Add the equations 11.9 They are circles of radii
along the three axis and simplify.
r = 1.4R (50%)
r = 3.1R (10%)
10.5
r = lOR (1 %)
~~ lA- ~~ IB
Streamlines:
azljl 1/14 + 1/!s- 21/11
r = e(Q/K)(O-Oo)
ox 2 a a2
Q K Similarly, o2 1/J/oy 2 is determined. Since '\1 2 1/J = 0, one finds 1/1 1.
1/1=-IJ--Inr
2n 2n
11.17 x = C cosh <P cos 1/J; y = C sinh <P sin 1/J. Equipotential lines ( =
11.12 Take the potential function for a source and a sink of same strength
constant):
apart by a distance 2a such as
xz yz
Q ~-~--c+ =I
<P =-(In r2 - In rd C2 cosh 2 <P C2 sinh 2 <P
2n
Streamlines (1/1 = constant):
where r 1 and r 2 are measured from source and sink, respectively. Insert
the relationships (see figure below):
ri = r 2 + a2 - 2ar cos (}
Foci: (O,C) and (0,-C).
r~ = r2 + a 2 + 2ar cos (}
Let 2a(Q/2n) = K and take the limit when a tends to zero.
Chapter 12
12.2 F = pQ
2
[~ - ~ (R 0 ) 2 - In ~]
n:Rf, 2 2 R R0
Energy:
Total force
vz vz
X = -J 0
2
pR cos (} diJ = 0
~ + !!.!._ + h 1 = _!. + Pz + h2 + llH
2g pg 2g pg
Y=-- pur J
"
2
sin 2 1JdiJ=-pUr
Momentum:
n o
Q rl
11.15 <P = -In- + Ur cos(}
2n r 2 First case:
Shape:
1/J = 0 gives
(h can sometimes be neglected by comparison with p/pg)
Answers to Selected Problems
F = fI
z p(x)
-sinrxdx
pg
PaQa(V, - V,)
Third case:
P Pa Chapter 13
pg pg
13.1 o* = o/2, 0/3; e = o/6, 2o/15; o** = o/4 ...
The sum of external forces due to atmospheric pressure equals zero;
so all the terms p/ pg disappear and F = 0. 13.2 Since 1/1 = (vxU 0 ) 112 f(Yf), '7 = y(U 0 /vx) 112
12.7 pQV = pgzA (external force), and since pQV = p(2gz)AC" C, =!.
z2 h2
12.8 pg 2- pg 2 = pq[V- f(rx)]
13.3 A 0 = 0, A 1 = 0, A 3 = 0, A4 = 0, A 5 = -!AL A 6 = 0, A7 = 0,
f(rx) ~ /2g(Zw + H- h)] 112 sin IX A8 = -JfAt A 9 = 0, A 10 = 0,
Inserting q = f(H), the function z = f(H) is obtained. 1 375 4
A 11 --~-A 2
12.9 F = pQ(V- U); power of the jet: pQ(V 2 /2). Transmitted power: - 8 11! '
pQ(V- U)U. In the case of the bucket, F = 2pQ(V- U).
13.4 y = 0, u= 0, o2 u/oy 2 = 0; y = o, u = U0 , oujoy = 0, o2 u/oy 2 = 0.
12.10 In the first case the pressure at the bottom of the vertical wall is
pgy 2 while in the second case it is pg(y 1 + h) depending upon the exact a 0 = 0, a 2 = 0, a 1 = 2(U 0 /0), a 3 =- 2U o/0 3 , a 4 = U 0 /0 4 .
location of the jump with respect to the bottom drop.
12.15 (a) Given V, wind velocity, D 1 the diameter of the windmill, A the
r0 = ou
pv- I 2pvU 0
oy FO 0
cross section of the boat and, C n its drag coefficient. The
unknowns are D 2 the diameter of the propeller, V, the velocity 13.5 r-f"+1=o
of the jet generated by the propeller. And the equations are
(Pais the density of the air and Pw that of the water). Multiplying this equation by f" and integrating,
(2 + f)l/2 (2)'12]
and '1 = (2) 112[ tanh- 1 (3)112 - tanh- 1 J
Finally,
The boat can go upwind if D 2 is such that this system of equa- ou ou 2 ouv I op o2 u
-+-+-=---+ V-.
tions is verified, and Vw and V, are subsequently determined. ot ox oy P ox oy2
Answers to Selected Problems
13.8 1. -d
dx
s 2 dy -
u u 0 -d
dx
s Ll dy 8u(O)
= - v- - + gii
h = R{ln [b + (b 2 - R 2) 112 ] -In R}
0 0 ay
2. a0 = 0, a 1 = 2, a 2 = - I
8 d 2 d
Chapter 16
2v
- - (2gxii) - - 2g(x) 112 - [(x) 112 ii] = - - (2gx) 112 + gii
15 dx 3 dx ii 16.1 Streamlines:
[U o = (2gx)tl2]
k sinh m(d + z)
3. f3 = 3(2)112 [v/(g)t/2]112 - cos mx = constant
m sinh md
(
3Q )4/3 Isobars:
4. Xo = 2(2g)!12f3
cosh m(d + z 0 )
(
3Q )1/3 z = -a
cosh md
cos (kt - mx)
5. Yo = f3 2(2g)!12f3
6. -
d s Ll
2
dy = -
au(O)
v- - + gii
16.2 Streamlines:
k sinh m(d + z) .
dx 0 i!y - sm mx = constant
m sinh md
Isobars:
cosh m(d + z 0 ) .
5g I [(ii* - ii 0 ) 2( ii* 2 + ii*ii + ;; 2 )] z = - 2a cos mx sm kt
8 6Qz (x - Xo) = 6/i* 2 In ii* - ii ii* 2 + ii*iio + ii5 cosh md
3vQ) 1
/3 3x 2 a2 2n (T )
9. ii3= ( - IJ =
-
a3 sinJ; t T~ = 31/2
g
Answers to Selected Problems
p H cosh m(d + z) . H
16.6 ~ = -z +- sm (kt- mx) 16.16 Ap = pg - - -
pg 2 cosh md cosh md
16.8 Y- Yo
---=-tanh m(d
X - Xo
+ z 0 ) cot mx 0 16.17 1. p= Pa- A oo2;, ooYf ~- oa4>1 ,
X t Z FO
16.12 The maximum pressure on the vertical wall is approximated by a 2. Insert 4> from Table 16-2 into the free surface condition, since
linear distribution between the elevation d + H where the pressure d(cosh a) = sinh ada, d(sinh a) = cosh ada. One obtains
is zero, and the bottom where the pressure is pg{d + H[cosh (2nd/L)] - 1}.
The underpressure acting on the vertical breakwater is assumed to be C2 = (~)2 = (gL + ~ 2n) tanh 2nd
distributed between this latter expression and pgd on the harbor side T 2n p L L
of the breakwater. The pressure on the harbor side is hydrostatic. The
overturning momentum and bottom stress are then determined as in a Water: Cmin = 23.1 em and L = 1.71 em; Mercury: Cmin = 19.4
gravity dam. . em and L = 1.202 em
3. When T tends to zero, C tends to infinity. When T tends to
16.14 Distance between orthogonals:
infinity, C tends to gT/2n. The capillary effect becomes negligible
b0 cos IX 0 when T > 0.3 sec in water.
bb cos l)(b
IX
2ndb
= - - , t = tanh IX, s = sinh 21X
1 84>
grad [ --+-+gz+K4>
p
e at P
J =0
Lb
Then At the free surface where z = Yf,
21X)1/2 H ( 1 - sin2 )1/4
0.14t512 ( 1 + ~ = ~
IX
0 a, 1 a4>
s L0 1 - t 2 sin 2 IXo
at e az
which gives db/ Lb as a function of H 0 , L 0 , and IX 0 . So Lb is obtained from
Lb/L 0 = t and db and finally IXb.
Chapter 17
cosh m(d + z)
16.15 1. Ap = pgH - - - -
cosh md 17.1 l. A = 4.24 ft (1.29 m)
cosh md 2. 'lmax = 15.24 ft (4.65 m)
2. K = ----:,.---:-:-------.,. 'I min = 6.76 ft (2.06 m)
cosh m(d + z 0 )
3. Ap = 412 lb/ft 2 (2009 kg/m 2)
3. H = 6.5 ft (1.98 m) 4. 37,800 lb/ft run at 13.7 ft above the base (56.179 kg/m at 4.17 m)
Answers to Selected Problems
p(y) = 2p(x) dy
dx
p = pg[z + _I_]
cosh kd
= 0 for y < 0
From the general theory of Fourier analysis, this integral only exists
for any specified value of m when R(r) has sinusoidal components con-
taining the argument mwr. Since R(r) is an even function, the assumed
Fourier series
A.6 b = 2a,
B
Backwater curves, 187-188
Banki, experience of, 110
Barre de Saint Venant, 258
equation of, 270
Basic equations in fluid mechanics, 9
Basic flow patterns, 124
Beaches (scale model), 298
Bergeron, paradox of, 143
Bernoulli equation, 23, 101
rotational flow, 61
stream tube, 109
two forms of, 109
water waves, 213
Bessel function, 225
Blasius equation, 157
Blasius theory, 179
Body forces, 51
Index
Borda mouthpiece, !50 Cofferdam (scale model), 298 Damping factor, turbulent boundary layer, 230
Bore, 262, 264 Comparison between different kinds of Damping
Bottom friction, for water waves, 229 motion, 96 of gravity wave, 65
Bottom outlet, scale model, 298 Complex numbers, 133 viscous wave, 295
Boundary conditions in water waves, 221 wave, 229
definition, II Compression (continuity), 33 Darcy law, 92
infinity, at, 13 Conditions of similitude, 288 Darcy (Weisbach equation), 167
importance of on methods of solutions, 129 Conductivity, porous medium, 93 Dead zone, 172
water waves, for, 213 Cone of diffusion, in pressure medium, 89 Dean stream function wave theory, 243
Boundary layer, 23 Conformal mapping, 133 Deep water waves, flow pattern, 7
approximation, 43 Conjugate depth, in hydraulic jump, 189 Deformation of elementary fluid
definition, !54 Continuity particle, 16, 18
momentum integral equation, !59 equation Delta function, 278
over a flat plate, unsteady flow, 161 estuary, river, 257 Densimetric Froude number, 297
separation, 172 general case, 31 Density effects (scale model), 295
thickness, 155 three-dimensional long wave, 256 Depth in uniform flow, 182
turbulent motion, 163 pipe, in a, 31 Derivatives of a vector, 46
Boussinesq equation for water waves, 259 potential function, and, 120 Diagram, wave refraction, 231
Boussinesq theory principle, 31, I 0 Diaphram, 110
of solitary waves, 243 relationship Diffusion equation, 254
of turbulent flow, 84 porous medium, 90 Dilatational deformation, 18
Breakwater turbulent flow, 75 coefficient, 27
effect of end, 232 Convective Dirac delta function, 278
scale model, 298 acceleration, 39 Discharge, as functions of stream
stability, 295 forces, approximations, 42 function, 118
inertia, mathematical expression, 40 Displacement thickness of boundary
Convergent channel, boundary layer in, 180 layer, 155
c Converging wave, 225
Conveyance of a channel, 183
Dissipative flow, 24
Dissipation function
Capillary Coridis components, 257 turbulent flow, 87
coefficient, 237 Coriolis acceleration, 44 viscous flow, 57
effects Correlation coefficients, 78 Dissipation thickness, boundary layer, 179
porous medium, 92 Cosh, 234 Distortion, scale model, 292
water waves, in, 294 Criteria of similitude, 288 Diverging wave, 225
forces, 52 Critical depth, 184--185 Domain of dependance, 264
waves, 237, 313 Critical slope, 186--188 Domain of influence, 264
Cartesian coordinate system, 4 Crossing of wave orthogonals, 231 Doublet, 124, 135
Cauchy--Poisson condition, 214 Curvature Drag coefficient (CD), 178
Channel, 182 flow, 191 cylinder, 17 5
unity, 188 effect of, 259 sphere, 173
Characteristics, 262 free surface, 224 water wave, in, 296
Chezy coefficient, 182 Cylinder Drag
Chezy formula, 182 added mass, 177 immersed body, on, 171
Circular hydraulic jump, 192 drag coefficient, 175 plate, of a, !58
Circular tank, wave agitation in a, 226 flow past a, 124 turbulent boundary layer, 166
Circulation of velocity, 124, 126 wave force on, 297 Du Buat paradox, 82
flow past a cylinder with, 134 Cylindrical system, 306--307 Dummy index, 60
Clapotis, 199 Cylindrical wave motion, 224 Dupuit approximation, 254
nonlinear, 255
Cnoidal wave, 201, 206, 239
Coaxial cylinders, 30 D
Coefficient(s) Dam
E
of dilatational deformation, 27 breaking, 262 Economic considerations in scale model, 287
of rotation, 27 break problems, 262-263 Efficiency of hydraulic jump, 192
of shear deformation, 27 effect of, 112 Ekman spiral, 70, 308
Index
Elasticity, similitude of, 298 Fluctuating components of motions, 71, 72 Harmonic solutions, 215
Elementary particle of fluids (definition), 3 Fluctuation velocity, 72 Harmonic wave, definition, 198
Elevation head, 104 Forced vortex, 20, 21, 125 Harbor agitation-Scale model, 298
Emptying a basin, 42 Form drag, 172 Head, hydraulic, 104
Energy Formulary Head loss
change, 104 progressive linear wave, 235 definition, II 0
density spectrum, 278 second-order approximation, 251 free surface flow, in, 112
dissipation, 24 standing wave, 236 hydraulic jump, in, 190
flux,226 water waves, 234 pipe line, in a, 87
in water waves, 235 Formulation of a surface wave similitude of, 291
internal forces, and, 138 problem, 212-214 singularity, at, Ill
per wave length, 226 Fourier analysis, 275-277 unsteady motion, 114
spectrum at sea, 280 Fourier transform, 279 Helmholtz equation, 216
Enlargement, sudden, 145 Free surface High-velocity flow, 132
Equation of state, II boundary conditions, II Horizontal motion (nearly), 48
Equipotential line, surface, 120 condition in linear harmonic wave, 217 Hydraulic conducting, porous medium, 93
Erfc functions, 162 disturbance, 186 Hydraulic jump, 141, 188
Estuary-continuity equation, 257 elevation in water waves, 248 efficiency, 192
Euler equation, 59 flow, singular head loss, 112 equation, 142
vectorial notation, 59 Free turbulence, 68 as function of Froude number, !50
Eulerian method, definition, 8 Free vortex, 21, 107, 125-127 similitude of energy dissipation, 290
Expansion, continuity, 33 Frequency, reduced, 70 slope, on a, 146
Exact solutions- Navier-Stokes equation, 63 Friction drag, 172 sudden deepening, 145, !51
Explosion, wave generated by, 208 Friction forces, 4 tunnel, in a, 142
External forces, 51, 137 flow with, 130 Hydraulic radius, 182
Froude number, 70 Hydrostatics, 52, 107
F Froude similitude, 289 law, 103
Hyperbolic function, 234
Fahrenheit degree, II
Families of water waves, 205
G
Fast Fourier transform, 279 I
Fender (similitude), 298 Galilean transformation, 6
Filter, wave, 237 Gallery (similitude), 291 Ideal fluid, definition, 4
Finite amplitude wave(s), 239 Generalized Bernoulli equation, 107 Impulsive motion, of or over a flat plate, 162
introduction, 208 Generalized Fr.oude similitude, 289 Incompressible fluid, continuity, two
Flood wave, 258 Gentle slope, 188 dimensional, 31
Flow Geostrophic Induced mass, 174
around cylinder, 172 acceleration, 44 Inertial coefficient (CM), 178
around sphere, 114, 172 forces, 52 Intake, 144, 146
between parallel planes, turbulent, 83 inertial components, 45, 48 Internal forces, 51, 137
between two parallel planes, 69 Gerstner wave theory, 23, 244 Inertial forces, 37
control, in open channel, 186 Gertoner wave, 65 vectorial notation, 42
in pipe, 166 Goda wave theory, 243 Inversen modulus, 178, 296
in porous medium, 65 Gradient of pressure, 53 Irrotational finite-amplitude wave
in wall angle, 134 Gradually varied flow, 186-187 calculation, 249
net, 130 Graphical method, for addition of flow Irrotational flow(s), 24
free surface, with, 132 patterns, 127 Bernoulli equation, I 02
porous medium in, 132 Gravity forces, 52 continuity, 35
past a cylinder, 128 Group velocity, 226, 227, 228 definition, 16
conformal representation, 134 without friction, 147
pattern(s), 116 Irrotational motion
combination of, 127 H steady, 104
in progressive periodic wave, 219-223 turbulent flow, 81
in standing periodic wave, 219-223 Hankel function, 225 Isothermal flow, II
under pressure, 128, 131 Harmonic analysis, 275 Isotropic turbulence, 78
Index
Kelvin theorem, 22
Kelvin waves, 208 N
Keulegan and Patterson, 243 M Navier-Stokes equation, 59
Kinetic energy in water waves, 226 and momentum theorem, 138
Korteweg and de Vries Manifold, 148
Manning coefficient, 183 numerical method, 66
parameter, 204 tensorial notation, 60
theory, 240 Manning formula, 183
Markers and cells (MAC) method, 66 vectorial equation, 61-62
Mass transport, 240 Nekrassov, 241
formulary, 251 Net (flow), 130
L Mathematical treatment of basic Newtonian
Lacey, law of, 293 equations, 100 equation, 37
Lagrangian McCowan (theory of solitary wave), 243 first law, 10
coordinates for water waves, 242 Mean forces fluid, 4
equation, 60 calculation of, 73 second law, 10
method, definition, 7 porous medium, 93 Nikuradse resistance coefficient, 171
system of coordinates (continuity), 35 porous medium, in, 90 Node, 220-224
Laitone wave theory, 243 turbulent motion, 72 Nonisotropic flow in porous medium, 89, 93
Lame components, 55 Mean motion Nonlinear effects
turbulent flow, 77 in a stream tube, I 08 on probability distribution, 283
Laminar boundary layer over a flat plate, turbulent flow, 71, 72 on wave refraction, 231
steady flow, 156 Method of characteristics, 262 Nonlinearities
Laminar flow in pipe, 166 Michell limit on wave breaking, 200 on spectral operations, 284
Laminar flow, stability of a, 66 Midlevel elevation in progressive wave, in water wave, introduction, 204
Laplace equation, 35, 213 second order, 251 Normal
Law of Darcy, 92, 93, 94 Midwater level in a standing wave, 255 depth, 182-183
Levi Civita, 241 Mixing length, 85, 86 discharge, 183
Limit of validity of Law of Darcy, 94 Momentum forces, 56
Limit wave steepness, 296 change Notation, 301
effect of vorticity, 24 7 with respect to space, 139 Numerical methods of solution
Linear deformation, 18 with respect to time, 139 direct approach, 265
acceleration, 39 vs. energy, 147 long waves, 260
Linear forces, turbulent flow, 73 equation, 10, 59 Navier-Stokes equation, 66
Linear long wave theory, 242, 260 along a streamline, 101
Linear wave theory, 212 general form, 61-63
introduction, 203 long wave, for, 257 0
Linearization, 43 water waves, for, 213
Local acceleration, 38 integral equation for boundary Open-channel hydraulics, 182
Local inertia forces, mathematical layer, 159-160 Orbits
expression, 40 theorem in progressive wave, 219-223
Locks, 115 general, 137 water waves, 235
hydraulic, 309 mathematical documentation, 139 Orthogonal wave, 230
Logarithmic law for velocity stream tube, applied to a, 140 Oscillatory waves, definition, 198
distribution, 168 unsteady flow, applied to, 145 Oseen theory, 44
Long model, 291 thickness of boundary layer, 155 Outfall, 297
Index